Keywords: Machine Learning in modeling, estimation, and control
Abstract: Many approaches to machine learning have struggled with applications that possess complex process dynamics. In contrast, human intelligence is adapted, and - - arguably - built to deal with complex dynamics. The current theory holds that human brain achieves that by constantly rebuilding a model of the world based on the feedback it receives. I will describe an approach to machine learning of dynamical systems based on Koopman Operator Theory (KOT) that also produces generative, predictive, context-aware models amenable to (feedback) control applications. KOT has deep mathematical roots and I will discuss its basic tenets. I will also present computational methods that enable lean computation. A number of examples will be discussed, including use in fluid dynamics, power grid dynamics, network security, soft robotics, and game dynamics.

Keywords: Manufacturing Systems, Control Design, Mining, Mineral and Metal Processing
Abstract: Strip metal manufacturing using conventional rolling processes is known to be energy and carbon emission intensive as it involves multiple hot and cold rolling stages to produce coils of thin metal strips. A novel clean energy alternative to conventional rolling technology is metal peeling which can produce a continuous metal strip by machining a thin surface layer directly from a rotating feedstock and coupling the produced strip into a roll-to-roll system for transport, finishing, and coiling. However, integrating metal peeling with a tensioning and coiling roll-to-roll system presents several challenges. For example: (1) strip thickness is a function of the downstream tension applied on the peeled strip as it is transported on rollers to the coiler; (2) workpiece characteristics and upstream peeling parameters affect the strip properties, such as surface quality and geometry; and (3) capturing the initial peeled strip, splicing, and threading it into the roll-to-roll system. Governing equations for strip tension and transport speed are available for unwinding from a roll, transport through finishing processes, and winding on to a coil. In contrast, there are no available models that describe the peeling process coupled with transport of the peeled strip under tension through the roll-to-roll system. In addition to the modeling challenges, there are control challenges associated with simultaneously controlling strip tension and strip transport speed, which requires development of a model-based control system to address the interaction between the peeling process parameters and transport variables. Specifically, inhomogeneous workpiece characteristics manifest as thickness variations and disturbing forces during peeling, making tension and speed regulation even more challenging. Thus, another control requirement is the ability to mitigate disturbances on the strip due to the peeling process and its transport. This presentation will describe (1) the challenges associated with modeling and control of this new manufacturing system; and (2) provide possible approaches for developing the governing equations for key process and transport variables and control systems to regulate them.

Keywords: Manufacturing Systems, Control Applications
Abstract: Traditional thermographic feedback control methods for Additive Manufacturing often cast 2D thermal profile data to a 1D feedback metric, such as meltpool area or width, by thresholding the data and returning the number of surviving pixels or greatest span in a specified direction. We extend these methods to encode additional spatial information into the feedback signal by introducing a 2D elementwise weighting matrix, where each weight corresponds to a linear combination of the observed deposition modes. Consequently, by replacing morphology-agnostic metrics such as meltpool area with a spatially-informed process variable, both feedback sensitivity and authority over deposition morphology were anticipated to improve. The most recent closed-loop performance results are presented here, as evaluated using a Laser-Wire Additive Manufacturing process utilizing a Soda-Lime glass filament feedstock and weights identified via training a neural network on thermographic process data.

Keywords: Control Applications, Manufacturing Systems, Robotics
Abstract: Grinding processes in manufacturing are typically performed by skilled technicians, which exposes them to airborne metal dust and leads to poor working conditions. Furthermore, since the technicians perform the grinding process manually, the part consistency and quality cannot be guaranteed. By automating the grinding process using a robot, the poor working conditions and part inconsistency can be mitigated; however, other issues arise with robotic grinding. For example, material removal can be affected by variations in the initial conditions of the part and grinding process, such as the amount of slag on a cast part. Additionally, a robot may be unable to perform the grinding process at the desired depth and accuracy on the first pass due to the low rigidity of the grinding disk. While some of these factors can be mitigated using standard force control methodologies, these methods do not provide in-situ measurements of the part’s accuracy. This work presents an initial theoretical and experimental iterative learning process framework in which the ground part is scanned using a laser radar and a new path is calculated to correct for errors from the previous iterations. This control method increases the part accuracy compared to the open loop process.

Keywords: Modelling, Identification and Signal Processing, Manufacturing Systems, Human-Machine and Human-Robot Systems
Abstract: Humans and machines provide the foundation for decision-making and action within Smart Manufacturing systems. While modeling machines based on real-time data and knowledge from subject matter experts has been widely explored, modeling humans has not. Modeling humans presents significant challenges due to the strong heterogeneity of individuals and the stochasticity that may be introduced through individual decisions to events. However, we hypothesize that human behavior within a controlled environment and task will lie within a predictable scope that can give insight into future behavioral states of humans (e.g., fatigue). Human fatigue affects workers' productivity, safety, and performance in manufacturing systems. We define fatigue as measurable tiredness resulting from physical exertion. If we can predict when a human worker will become fatigued, breaks or task changes can be preemptively incorporated into system planning to maintain desired levels of productivity and safety. In this paper, we propose a modeling structure to describe fatigue in humans performing repetitive, manufacturing-like tasks based on task context and physiological sensor data using first order modeling techniques. A case study, involving two participants, where the participants complete a repetitive cable assembly task provides a demonstration of the model development. A trial in this case study includes both working and resting periods, where participants wore wearable weights of varying amounts in order to provide variation in task context. The derived first order models demonstrated how differently weighted tasks resulted in first order responses with varying time constants as a function of context. Models across participants showed similar performance for individually tuned parameters versus a nominal model.

Keywords: Modelling, Identification and Signal Processing, Modeling and Validation, Control Design
Abstract: Solid-state batteries are a promising next-generation technology to meet the ever-growing demand for higher-performing and safer batteries with higher energy density, reduced flammability, and faster charging capabilities. However, the material interfaces are prone to severe and rapid degradation in the form of void formation and dendritic growth, resulting in performance losses or failure of the cell. Typically, battery health management for new battery technologies entails the use of physics-based modeling of the electro-chemical processes, which can be computationally complex and difficult for real-time health monitoring. Additionally, even with the computational cost of simulating battery health degradation in high-fidelity, statistical variation due to manufacturing processes will still present model inaccuracies in application. Approaching the problem with controls methodology provides an avenue for real-time monitoring and evaluation of solid-state battery health at the material interfaces. This project introduces a controls-oriented methodology for data-driven system identification and design evaluation for solid-state batteries. The complex electro-chemical processes governed by partial differential equations are reduced as a linear parameter varying model defined by ordinary differential equations. The reduced-order, data-driven model opens the door for computationally efficient real-time estimation and prediction, as well as closed-loop feedback control of solid-state interface degradation.

Keywords: Machine Learning in modeling, estimation, and control, Manufacturing Systems, Adaptive and Learning Systems
Abstract: Temperature distribution during the laser powder bed fusion (LPBF) additive manufacturing process can largely affect the quality of the printed parts. To improve the print quality, such as mitigating the deformation and defects caused by non-uniform temperature distribution, model-based control techniques can be used to optimize the scanning strategy to achieve desired temperature distribution. However, existing temperature prediction models are either inaccurate or computationally expensive, harming the controller performance or even inapplicable to controllers. To design an accurate model that is suitable for model-based controllers, the authors propose a linear hybrid (i.e., physics-guided data-driven) model, which cascades a linear data-driven model with a physics-based model that is based on a finite difference method (FDM) thermal diffusion model. The linear data-driven model aims to correct the physics-based model prediction by learning from past measurements. It is designed in a linear fashion to be easily applied to controllers and also uses physics-informed features to enhance its interpretability. From simulation, the proposed linear hybrid model is shown to be able to capture the unmodeled dynamics and significantly improve the temperature prediction accuracy compared to the physics-based FDM model.

Keywords: Motion and Vibration Control, Robotics, Mechatronic Systems
Abstract: The inherent lightweight nature of collaborative robots renders them susceptible to vibrations when tasked with rapid movements and substantial loads. This research proposes a technique for controlling vibrations in 6-degrees-of-freedom collaborative robots, specifically targeting the minimization of residual vibrations at the end-effector. To achieve this, a time-varying filtered B-splines (FBS) approach will be employed to suppress residual vibrations in a UR5e industrial robot operating at high speeds, while ensuring minimal impact on tracking performance. The experimental results are awaited to demonstrate the effectiveness of this method, even when confronted with continuously changing robot dynamics during operation. A Robust Time-varying Input-Shaping (ZVD) will serve as the residual vibration reduction benchmark while Time-invariant FBS will be used as a reference to evaluate tracking performance. It is expected that the Time-Varying FBS approach will yield comparable vibration reduction results to Input-Shaping, without compromising tracking performance or increasing trajectory time. Furthermore, the proposed method should present better tracking performance than Time-invariant FBS.

Keywords: Manufacturing Systems, Estimation, Mechatronic Systems
Abstract: Compared to conventional electronics fabrication techniques, printed electronics have the advantages of reducing complexity and cost, while increasing material compatibility and design flexibility. Electrohydrodynamic jet (E-jet) printing is a 3D printing process with high-resolution capabilities, and it has shown great potential in fabricating micron-scale electrical components with metal nanoparticle inks. While examples of E-jet printed conductive interconnects are demonstrated and the sintering mechanism of metal nanoparticles are well studied, a knowledge gap exists in our understanding of the electrical performance of E-jet printed interconnects and the effect of sintering to them specifically. This work highlights the resistivity characterization of near micron-scale (1-3 μm) printed interconnects, and the derivation of models that relates interconnect resistivity to sintering conditions and shape metrics which enable resistivity estimation and process optimization.

Keywords: Robotics, Manufacturing Systems, Control Applications
Abstract: Within the past decade, a variety of Metrology-in-the-Loop control strategies have been developed to improve the accuracy of industrial robots for manufacturing applications. These strategies typically incorporate external metrology systems, such as laser trackers, in traditional motion control architectures around the robot’s existing control system. A challenge with these strategies arises from the often significant and delayed response of the robot to external corrections, which limits the achievable bandwidth and robustness of the combined control system. This poster summarizes recent efforts to develop a new Metrology-in-the-Loop control strategy that utilizes a disturbance observer architecture to improving robot accuracy through real-time kinematic error compensation. The proposed architecture operates harmoniously with the robot’s existing motion control loop, allowing for more robust accuracy improvements of the robot, despite the robot’s delayed response to external corrections. In addition to the control system architecture, this poster will also present the filtering strategy used to condition the kinematic error signal and the results of several experiments conducted to evaluate the performance of the Metrology-in-the-Loop control system.

Keywords: Adaptive and Learning Systems, Control Applications, Distributed Parameter Systems
Abstract: Single Point Incremental Forming (SPIF) is a freeform manufacturing technology for the rapid production of sheet metal parts. In SPIF, a single, stylus-like tool traces the contour of the CAD geometry in increasing depths. As the tool contacts the clamped workpiece, highly concentrated force occurs at the tooltip, resulting in plastic deformation of the part. Despite its potential for small batch size applications, SPIF has failed to be widely adopted in many industrial and medical applications due to low geometric accuracy, especially for complex part features. Iterative Learning Control (ILC) feedback may be used to correct the part’s geometric error, however, the SPIF process provides several unique challenges to ILC. The continuous change in material properties from cold-working and the sheet geometry throughout the process make it difficult to predict how toolpath corrections will affect the sheet surface. Further, the lack of negative control authority due to the tool’s inability to pull the material, makes it impossible to correct features which have formed beyond the desired depth. This poster presents the latest results for a design methodology used to synthesize spatial learning filters for use in the ILC algorithm in order to overcome the challenges unique to SPIF. Finally, experimental results are presented to evaluate the system performance.

Keywords: Robotics, Machine Learning in modeling, estimation, and control, Adaptive and Learning Systems
Abstract: Slip detection in object handling is crucial and traditionally relies on visual cues. However, for optimal performance, artificial tactile sensing is needed, especially with unfamiliar objects. This study introduces a real-time, physics-informed, data-driven method for continuous slip detection using the GelSight Mini optical tactile sensor. The sensor's inhomogeneity during slip events helps create unique features and recasts slip detection as a classification task. Tested on ten diverse objects, the best model achieved 99% accuracy. The work's practical use was demonstrated in a dynamic robotic manipulation task incorporating real-time slip detection and prevention.

Keywords: Manufacturing Systems, Machine Learning in modeling, estimation, and control
Abstract: The growing demand for customization in the manufacturing of functional devices has propelled the development of additive manufacturing methods that can be tailored to meet the specific end-user requirements while ensuring a controlled process behavior and performance characteristics. However, achieving a robust, repeatable, and predictable behavior for these processes is a challenge, leading to the current predominantly open-loop operation of AM processes. The focus of this poster is on the development of an algorithm for control of an extrusion-based additive manufacturing system leveraging a physics-informed data-driven modeling framework. We will present an understanding of the modeling requirements and a preliminary learning-based control framework to enhance the performance of the final printed device. This learning‐based process control method incorporates two architectures: (1) an internal learning of process fidelity and robustness, which incorporates the process dynamics using a physics-informed data-driven model and (2) an outer learning, which incorporates the part-performance dynamics for decision making given a design and the part functionality. This framework will also provide insights towards the transfer of knowledge we obtain from one pattern to the next device structure or material.

Keywords: Robotics, Manufacturing Systems, Machine Learning in modeling, estimation, and control
Abstract: This poster focuses on a robotic system for surface inspection, and the development of a collaborative robotic manipulation intelligence. We discuss the proposed system, results of experiments in defect classification on important engineering materials, and demonstrate the tools and novel control methods designed to perform subtasks involved in robotic surface inspection, including automated waypoint generation from a surface model of the part, intelligent path planning between imaging poses, and rejection of environmental disturbances such as illuminance variation across the part surface, motion-blur, and out-of-focus regions.

Keywords: Modelling, Identification and Signal Processing, Nonlinear Control Systems
Abstract: Compared to first-principles models, data-driven modeling approaches require less knowledge about a given system. However, there are still many challenges with applying common machine learning-based techniques for modeling dynamical processes. In this work, we propose a similarity-based modeling method for nonlinear dynamical systems that makes predictions based on the similarity between a new sample and samples that are stored in an available dataset. Compared to conventional data-driven methods for modeling dynamical systems, our approach likens modeling of the dynamic system to interpolation within a dataset that represents the process dynamics. Therefore, we leverage knowledge from the pre-training process used in natural language modeling and image processing to develop a process model of a nonlinear dynamical system and compare its performance to other data-driven approaches.

Keywords: Modeling and Validation, Manufacturing Systems, Mechatronic Systems
Abstract: While powder bed fusion (PBF) additive manufacturing offers many advantages and exciting applications, its broader adoption is hindered by issues with reliability and variations during the manufacturing process. To address this, researchers have identified the importance of using both finite element modeling and control-oriented modeling to predict and improve the quality of printed parts. In this paper, we propose a novel control-oriented multi-track melt pool width model that utilizes the superposition principle to account for the complex thermal interactions that occur during PBF. We validate the effectiveness of the model by applying a finite element model of the thermal fields in PBF.

Keywords: Adaptive and Learning Systems, Control Applications
Abstract: Iterative learning control (ILC) and repetitive process control (RPC) are layer-to-layer corrective schemes which are particularly well-suited for additive manufacturing (AM). Direct energy deposition (DED) processes have unique error-propagation dynamics which are best compensated for with RPC structures. Frequently in DED builds, a non-constant reference toolpath is required for geometrical constraints (e.g. rocket nozzles, other aerospace applications); however, current control structures are usually limited to constant toolpaths. This work explores an adaptation to an existing interpolating RPC structure capable of accommodating a broader range of non-constant reference toolpaths with greater stability. Modified equations are developed for this structure and resulting specimens are presented. Specimens are compared to previous interpolating RPC implementations and performance metrics are discussed. Further applications for advancement are explored.

Keywords: Path Planning and Motion Control, Modeling and Validation, Estimation
Abstract: Mixed Integer Linear Programs (MILPs) are often used in the path planning of both ground and aerial vehicles. Such a formulation of the path planning problem requires a linear objective function and constraints, limiting the fidelity of the the tracking of vehicle states. One such state often used is the the charge level of the on board battery. High-fidelity battery state estimation requires nonlinear differential equations to be solved. This state estimation is vital in path planning to ensure flyable paths, however when using a linear path planning problem cannot implement these nonlinear models. Poor accuracy in battery estimation during the path planning runs the risk of the planned path being feasible by the estimation model but in reality will deplete the battery to a critical level, resulting in a dangerous planned path. To the end of higher accuracy battery estimation within a linear framework, we test a simple linear battery model which predicts the change in state-of-charge (SOC) of a battery given a power draw, time duration, and current SOC in the context of an a-priori path plannign problem. This context differentiates itself from real-time estimation. In ahead-of-time path planning, the changes to battery draw are often assumed as a series of constant power draws as opposed to rapidly changing power draw which may occur in real-time battery tracking and estimation. The linear battery model is presented and then tested against alternate models in both numerical and in experimental tests. Further, the effect of the proposed linear model on the time-to-solve a resource constrained shortest path problem is also evaluated, where two different algorithms are used to solve the path planning problem. It is seen that the linear model performs well in battery state estimation while remaining implementable in a Linear Program or MILP, with minimal effect on the time-to-solve. This provides what we consider to be a worthwhile trade-off in improved accuracy relative to increased time-to-solve.

Keywords: Path Planning and Motion Control, Unmanned Ground and Aerial Vehicles, Machine Learning in modeling, estimation, and control
Abstract: The aim of this paper is to explore potential of feedback control design based on a switching composition of two controllers. The paper considers two stochastic optimal controllers for the motion of an unmanned aerial vehicle (UAV) in the horizontal and vertical planes. We show that the two controllers, one for reaching a point in the horizontal plane and the other for reaching and keeping a desired altitude in the vertical plane, can be computed using Cartesian coordinates. To reach a desired waypoint in 3D, both controllers are necessary while the vehicle has to reach simultaneously the horizontal and vertical navigation goals. For this reason, we compute the expected time of each controller toward its goal in 2D. Then, we propose a switching rule that guarantees the simultaneous arrival of each controller to its 2D goal, which is equivalent to the vehicle reaching the 3D waypoint. Finally, we explore a possibility for improvements of the switching rule using reinforcement learning and an actor-critic neural network. The paper results are illustrated by numerical simulations.

Keywords: Unmanned Ground and Aerial Vehicles, Intelligent Autonomous Vehicles, Robotics
Abstract: Autonomous vehicle navigation and control require both planning and successful following of a desired path through an environment. This is often performed by a hierarchical framework that includes both a path planning algorithm and a path following controller. However, these two levels often operate with different modeling assumptions, constraints, and cost functions, leading to several challenges when integrated. To demonstrate the challenges that arise specifically when mapped obstacles represent regions that can be traversed but with varying traversal costs, this paper builds an intuitive control hierarchy consisting of an A* path planner, a Kalman filter to estimate traversal costs, and a model predictive controller. While each of these elements have been deeply studied individually, this paper identifies challenges that can arise from their coupled behavior and categories existing literature that addresses some of these challenges. Simulation results for a ground vehicle traversing regions of varying cost illustrate both the benefits and potential pitfalls of this hierarchical integration, motivating its further study for vehicle autonomy.

Keywords: Optimal Control, Path Planning and Motion Control, Power and Energy Systems
Abstract: Here we study an optimal control problem involving energy management of a hybrid-fuel Unmanned Aerial Vehicle (UAV). The planning problem for a hybrid-fuel platform involves determining the path while managing the energy resources, which includes a policy for power modality switching whenever applicable. The hybrid-fuel platform considered here involves a generator and battery pack combined in a series fashion as energy sources on-board a UAV. Also included in the problem are the noise restrictions, which place constraints on generator operation depending on the airspace location. These emulate possible restrictions on UAV noise that occur in military surveillance missions or in urban path planning, where the collective noise of many UAVs, some with combustion engines, may be restricted in certain areas or times of the day. We present a hybrid methodology which starts from an initial path and generator pattern obtained from a mixed integer linear program (MILP) solution. The generator pattern from the discrete solution is then refined in an optimal control framework with an objective to minimize fuel usage, while considering the nonlinear battery and generator dynamics and noise-restriction constraints. Optimal control problem is solved with a nonlinear program solver, IPOPT. Numerical results are presented and analyzed with varying path lengths and scenarios. This work aims to serve as an initial study of this hybrid-fuel UAV problem within an optimal control framework, which can be extended to refinement of both the generator pattern and the trajectory in tandem, while considering vehicle and power dynamics that are often ignored in discrete path planning solutions.

Keywords: Path Planning and Motion Control, Adaptive and Learning Systems, Robotics
Abstract: In this paper, we propose an algorithm and implementation for real-time optimized kinodynamic motion planning for aerial vehicles with unknown dynamics in crowded environments. A random-sampling space-filling tree is used for both planning and rapidly replanning a path through the environment. Then, continuous-time Q-learning is used to approximately solve the resulting finite-horizon optimal control problem online to optimally track the planned path. To facilitate the Q-learning, we propose an actor-critic structure with integral reinforcement learning to approximate the Hamilton-Jacobi-Bellman equation. The critic approximates the Q-function while the actor approximates the control policy. We demonstrate our approach on custom drone hardware in which all planning, learning, and control computations are conducted onboard in real-time.

Keywords: Machine Learning in modeling, estimation, and control, Optimal Control, Path Planning and Motion Control
Abstract: In this study, finite horizon constrained trajectory optimization is tackled by using Artificial Neural Network with an embedded subspace manifold. The resulting network takes advantage of the reduced dimension search space guided by a bio-inspired motion rule. The input nodes of the network are interpreted as collocation points over the time domain transcribed by a pseudospectral discretization method. The activation function for each node is the inverse of the dynamical system. The weights and biases to be optimized in the network are analogous to the parameters of the motion rule. The network is optimized during training by minimizing an augmented loss function where the constraints are considered penalties. The proposed method is simulated in a collision avoidance trajectory planning problem of a mobile robot with two driving wheels and an attitude slewing maneuver problem of an asymmetric rigid body spacecraft.

Keywords: Estimation, Transportation Systems, Modelling, Identification and Signal Processing
Abstract: We are pleased to present our IEEE Access journal paper.

Vehicle sideslip angle is a key state for lateral vehicle dynamics, but measuring it is expensive and unpractical. Still, knowledge of this state would be really valuable for vehicle control systems aimed at enhancing vehicle safety, to help to reduce worldwide fatal car accidents. This has motivated the research community to investigate techniques to estimate vehicle sideslip angle, which is still a challenging problem. One of the major issues is the need for accurate tyre model parameters, which are difficult to characterise and subject to change during vehicle operation. This paper proposes a new method for estimating vehicle sideslip angle using an Extended Kalman Filter. The main novelties are: i) the tyre behaviour is described using a Rational tyre model whose parameters are estimated and updated online to account for their variation due to e.g. tyre wear and environmental conditions affecting the tyre behaviour; ii) the proposed technique is compared with two other methods available in the literature by means of experimental tests on a heavyduty vehicle. Results show that: i) the proposed method effectively estimates vehicle sideslip angle with an error limited to 0.5 deg in standard driving conditions, and less than 1 deg for a high-speed run; ii) the tyre parameters are successfully updated online, contributing to outclassing estimation methods based on tyre models that are either excessively simple or with non-varying parameters.

Keywords: Machine Learning in modeling, estimation, and control, Aerospace
Abstract: We present a novel data-driven approach for prediction of the estimated time of arrival (ETA) of aircraft in the terminal area via the implementation of a Random Forest regression model. The model uses data fused from a number of sources (flight track, weather, flight plan information, etc.) and provides predictions for the remaining flight time for aircraft landing at Dallas/Fort Worth (DFW) International Airport. The predictions are made when the aircraft is at a distance of 200-miles from the airport. The results show that the model is able to predict estimated time of arrival to within ± 5 min for 90% of the flights in the test data with the mean absolute error being lower at 145 seconds. This paper covers the entire pipeline of data collection, pre-processing, setup and training of the ML model, and the results obtained for DFW.

Keywords: Mechatronic Systems, Modeling and Validation, Estimation
Abstract: A critical challenge for industrial Atomic Force Microscopes (AFM) is automatic selection of appropriate probes required for the scanning and measurement tasks (or sequences of tasks). This challenge can be addressed by specially assigned multi-probe cartridge holders with replaceable probes. Cantilevers with different spring constants and resonant frequencies are required to make specific AFM measurements. In this paper the classical PDE of lateral vibration of bars of variable cross-sections to model dynamics of the AFM cantilevers and calculate their spring constants is used. In addition, utilizing Rayleigh method, resonant frequency can be estimated. Using software based on these models, probe designer can interactively select appropriate geometric and material parameters to achieve desired spring constants and resonant frequencies of the cantilevers to be assembled in the multi-probe cartridge holders.

Keywords: Estimation, Robotics
Abstract: State estimation of hybrid dynamic systems, such as legged robots, is challenging because of the presence of non-smooth dynamics. This paper applies the Unscented Kalman Filter (UKF) state estimator and two novel hybrid extensions (HUKF) to a hybrid system, the simplest walking model. These estimators are identical far from the switching boundary, which partitions dynamic domains, but apply different time update algorithms at the switching boundary. (1) UKF permits sigma points to propagate through the system's hybrid dynamics. (2) HUKF-SPG (Sigma Point Generation) generates new sigma points when the weighted mean of the initial sigma points is on the switching boundary. (3) HUKF-SPT (Sigma Point Transformation) transforms the sigma points forward and backward in time through the system's hybrid dynamics only when the weighted mean of the initial sigma points is on the switching boundary. Results here shows that HUKF-SPG and HUKF-SPT have a lower absolute error but modestly more computations compared to UKF. A caveat of HUKF-SPT is it can only apply to conservative systems. HUKF-SPG is more general and could be applied to any hybrid system.

Keywords: Estimation, Multi-agent and Networked Systems, Unmanned Ground and Aerial Vehicles
Abstract: We consider a wind field estimation problem with multiple quadcopters. The wind field is assumed to affect the motion of the quadcopters in an additive fashion. Starting with a single quadcopter case, we first design an Extended Kalman filter (EKF) for constant and spatial-varying wind estimation. We next extend the EKF wind estimator for multiple quadcopters with directed connected communication graphs. To fuse the estimates of the wind field, we develop a sequential covariance intersection (SCI) method and a sequential weighted exponential product (SWEP) method for constant and spatially-varying wind fields. The effectiveness of the designed partial state fusion methods is validated and compared in simulations for various communication topologies with constant and Rankine wind models.

Keywords: Automotive Systems, Machine Learning in modeling, estimation, and control
Abstract: The NOx conversion efficiency of a combined Selective Catalytic Reduction and Ammonia Slip Catalyst (SCR-ASC) in a Diesel Aftertreatment (AT) system degrades with time. A novel model-informed data-driven On-Board Diagnostic (OBD) binary classification strategy is proposed in this paper to distinguish an End of Useful Life (EUL) SCR-ASC catalyst from Degreened (DG) ones. An optimized, supervised machine learning model was used for the classification with a calibrated single-cell 3-state Continuous Stirred Tank Reactor (CSTR) observer used for state estimation. Percentage of samples classified as EUL (%EUL) w.r.t. classification boundary of 50%, was used as an objective criterion of classification. The method resulted in 87.5% classification accuracy when tested on 8 day-files from 4 trucks (2 day-files per truck; 1 DG and 1 EUL) operating in real-world on-road conditions. Each day-file had ~86,000 samples of data. Mileage of the same truck was used as ground truth for classification. However, mileage across different trucks cannot be used for classification since the operating conditions would vary across trucks.

Keywords: Machine Learning in modeling, estimation, and control, Power and Energy Systems, Modeling and Validation
Abstract: Accurate modeling is crucial for the effective design and control of fuel cell stacks. Although physics-based models are widely used, data-driven methods such as the Koopman operator have not been fully explored for fuel cell modeling. In this paper, a Koopman-based approach is utilized to model the thermal dynamics of a 5 kW open cathode proton exchange membrane fuel cell stack. A physics-based model is used as the baseline for comparison. By varying the cooling fan rotational speed, the dynamics of the fuel cell stack were measured from the low load of near 0 kW to about 5 kW. Compared to experimental results, the errors of Koopman-based models are within 3%. Additionally, once given sufficient dimension, the Koopman-based model outperforms the physics-based method in both accuracy and computational time. These findings suggest the potential of the Koopman operator as an alternative approach for fuel cell stack modeling, contributing to the development of more accurate and efficient modeling methods for fuel cell systems, and facilitating the implementation of the linear optimal algorithms.

Keywords: Automotive Systems, Modelling, Identification and Signal Processing, Estimation
Abstract: The parameters of an active suspension need to be identified online, such that the suspension control system can be adapted to mechanical wear and load change. Recursive least squares and observer-based methods are frequently utilized to fulfill this purpose. However, they can yield slow parameter identification due to their asymptotic nature. We propose an algebraic identifier to estimate the parameters of an active suspension online, which does not maintain an asymptotic convergence phase. Simulation results demonstrate the effectiveness of the proposed algebraic approach.

Keywords: Control Applications, Optimal Control, Automotive Systems
Abstract: The primary barrier to implementing the RCCI mode in on-road vehicles is its limited operating range due to excessive pressure rise rates and complexity of combustion control during mode switching. The feasible operating range of the RCCI mode is only a subset of the speed-load range of conventional spark ignition (SI) engines. Therefore, a multi-mode engine concept operating in RCCI and SI modes can cover required engine operating range and can improve engine performance in terms of thermal efficiency and emissions. The goal of this study is to develop a model-based closed loop control of an SI-RCCI multi-mode engine. To this end, a gain scheduled model predictive controller (MPC) is developed for SI-RCCI-SI mode switching. A Kalman filter is designed for state estimations. The control architecture includes; a supervisory controller that determines the optimal operating mode based on the requested engine load and speed, and gain scheduled MPC and a Kalman filter. The controller performance is validated on a 2-liter multi-mode engine for SI-RCCI-SI mode switching at different engine loads (NMEP) and combustion phasings (CA50). The controller is capable of tracking NMEP and CA50 in both modes and during mode transitions while maintaining air-fuel ratio near stoichiometry in SI mode and constraining the maximum pressure rise rate (MPRR) below 8 bar/CAD in the RCCI mode.

Keywords: Intelligent Autonomous Vehicles, Automotive Systems, Machine Learning in modeling, estimation, and control
Abstract: Model uncertainties are considered in a learning-based control framework that combines control dependent barrier function (CDBF), time-varying control barrier function (TCBF), and control Lyapunov function (CLF). Tracking control is achieved by CLF, while safety-critical constraints during tracking are guaranteed by CDBF and TCBF. A reinforcement learning (RL) method is applied to jointly learn model uncertainties that related to CDBF, TCBF, and CLF. The learning-based framework eventually formulates a quadratic programming (QP) with different constraints of CDBF, TCBF and CLF involving model uncertainties. It is the first time to apply the proposed learning-based framework for safety-guaranteed tracking control of automated vehicles with uncertainties. The control performances are validated for two different single-lane change maneuvers via Simulink/CarSim® co-simulation and compared for the cases with and without learning. Moreover, the learning effects are discussed through explainable constraints in the QP formulation.

Keywords: Automotive Systems, Modeling and Validation, Optimal Control
Abstract: Automotive control algorithms have been moving from traditional, rule-based control algorithms to optimal, model-based control algorithms. While the model-based algorithms have shown to provide better control performance over a wide range of use cases with reduced calibration efforts, it is challenging to integrate multiple model-based controllers that are developed independently to address different control objectives. In this work, a nonlinear model predictive engine controller designed for torque tracking is integrated with a reference governor-based driveline controller designed to reduce vehicle drivability problems known as clunk and shuffle. The design of both controllers is discussed and their individual performance is demonstrated for real-world test conditions and realistic driving scenarios. Then, the integration between the two optimal controllers is discussed and a proof of concept use case showing the coordinated control between the two optimal controllers is presented. The results illustrate that without coordination between the two controllers neither of them are able to meet their original control objectives.

Keywords: Nonlinear Control Systems, Automotive Systems, Control Design
Abstract: Regenerative braking is one of the main advantages of electric propulsion systems. In such systems the vehicle brake controller has to prioritize safety while maximizing the recovered energy at all time. This paper proposes a two-step hierarchical brake controller for a dual-motor all-wheel-drive electric vehicle. In the first step a novel sliding mode controller (SMC) generates the total braking torque on each axle to independently control the slip on the front and rear wheels. In the second step the torque split controller assigns motor torques to maximize the recovered energy. The proposed SMC controller accounts for the non-linearities in vehicle dynamics and tire model, and considers the weight transfer due to vehicle deceleration. We will show in simulations that while the traditional SMC formulation is not effective during emergency braking scenarios, the proposed formulation successfully generates control commands to bring the vehicle to a stop position in a minimum distance. Furthermore, our controller is able to maintain both wheels in the stable slip region, even when starting from a locked position. The performance of the proposed controller is evaluated for an emergency braking scenario and on a segment of the US06 cycle.

Keywords: Linear Control Systems, Control Design, Estimation
Abstract: Dynamic mode decomposition with control (DMDc) has emerged in recent years as a powerful tool for data-driven system identification. Our work provides an extension to DMDc, namely an additional use case for the algorithm, that of input-delayed control systems with a time delay that is not necessarily commensurate with the sampling interval. Specifically, we propose a natural extension of DMDc to systems with a delayed input, which we coin dynamic mode decomposition with input-delayed Control (DMDidc). We prove that the method allows identification of an input-delayed linear time-invariant (LTI) system given sufficient input-output data. Moreover, we propose an extended predictive feedback control scheme that can accommodate delays that are a fraction of a discrete controllers sampling period. Additionally, we provide a numerical simulation to illustrate the benefits of fractional delay compensation. Our method extends DMDc to be a more practical choice for input-delayed system identification.

Keywords: Adaptive and Learning Systems, Control Design, Nonlinear Control Systems
Abstract: This article presents an adaptive artificial neural network-based control scheme that strategically integrates the attracting-manifold design and a smooth Lipschitz-constant projection operator. The essence of the scheme is elaborated through the design of a reference-command tracking control law of an nth-order single-input uncertain nonlinear system that can be transformed into the Brunovsky form. The method offers two major advantages. First, by employing the attracting-manifold design, it is possible to achieve asymptotic recovery of the ideal (deterministic) closed-loop dynamics. In other words, the perturbation caused by parametric uncertainties will be driven to zero asymptotically as a result of such a design, fostering a superior control performance. Second, the established smooth projection operator ensures the Lipschitz constant of the adaptive artificial neural network is bounded from above, thereby providing a certain degree of robustness against adversarial perturbations. The proposed method is validated through a numerical simulation example and compared with a standard certainty-equivalent neural-adaptive control method to demonstrate its superior performance.

Keywords: Assistive and Rehabilitation Robotics, Human-Machine and Human-Robot Systems, Control Applications
Abstract: Robotic exoskeletons for the hand are being explored to improve health, safety, and physical performance. However, much research effort is needed to establish reliable models of human behavior for effective human-robot interaction control. In this work, surface electromyography is used to measure and model muscle activity of healthy participants performing quasi-isometric and dynamic hand exercises. Non-negative matrix tri-factorization (NM3F) is used to extract hidden neuromuscular parameters encoded in spatial and temporal muscle synergies, which are used to estimate probabilistic linear models of intent, effort, and fatigue. This paper thereby presents steps toward reliable modeling of nonlinear time-varying hand neuromuscular dynamics for intuitive and robust human-robot interaction.

Keywords: Machine Learning in modeling, estimation, and control
Abstract: Data-driven predictive control has recently gained increasing attention, as it makes it possible to design constrained controls directly from a set of data, without requiring an intermediate identification step. In this paper, we focus on a Subspace Predictive Control (SPC) scheme, with the aim of clarifying the sensitivity of the final closed-loop performance to its main hyperparameters, namely the length of the past horizon and the regularization penalties. Moreover, by delving deep into the structural properties of the control problem formulation, we provide a set of guidelines for the choice of such hyperparameters. The effectiveness of the resulting overall tuning strategy is assessed on two benchmark examples.

Keywords: Manufacturing Systems, Robotics
Abstract: In manufacturing operations such as clamping and drilling of elastic structures, tool-workpiece normality must be maintained, and shear forces minimized to avoid tool or workpiece damage. The challenge is that the combined stiffness of a robot and workpiece, needed to control the robot-workpiece elastic interactions are often difficult to model and can vary due to geometry changes of the workpiece caused by large deformations and associated pose variations of the robot. The main contribution of this article is an algorithm (i) to learn the robot-workpiece stiffness relationship using a model-free data-based approach and (ii) to use it for applying desired forces and torques on the elastic structure. Moreover, comparative experiments with and without the data-based stiffness estimation show that clamping operating speed is increased by four times when using the stiffness estimation method while interaction forces and torques are kept within acceptable bounds.

Keywords: Automotive Systems, Estimation, Sensors and Actuators
Abstract: This paper presents a realtime recursive algorithm that can estimate, starting from an unknown pose, the mounting angles (roll, pitch and yaw) of an inertial sensor unit using accelerations and angular velocities. We analyze the use case of telematic boxes (E-Box) that are mounted on ground vehicles for safety reason (like E-call or automatic crash detection) or driving style monitoring. In order to work properly and record meaningful data, the box reference frame needs to be correctly aligned with the vehicle one. The proposed algorithm aligns the two reference frame online while the car is running throughout a series of filters and data point selection logics. Results show that the algorithm is robust with respect to any box mounting position or vehicle, with, on average, a convergence time of less than 20 minutes to the correct angles.

Keywords: Path Planning and Motion Control, Control Design
Abstract: Pure pursuit control is a commonly used method in the vehicle path-tracking problem. While widely used, it has some limitations, including the fact that its accuracy is very sensitive to the look ahead distance. A variety of different modifications to pure pursuit have been developed in an attempt to deal with this, and in this paper, an additional modification is proposed that combines speed and understeer gradient sensitivity with a curvature-based gain compensation based on target points on the path. Results are compared to the standard pure pursuit controller, demonstrating the improvements from this modification.

Keywords: Automotive Systems, Control Design, Control Applications
Abstract: In automotive control, torque vectoring enhances a vehicle dynamical characteristic by independently allocating wheel torques. Torque vectoring algorithms are often based on some form of closed-loop yaw rate tracking, whose reference is generated according to the driver's steering input. Thus, unless the vehicle is equipped with steer-by-wire, from the torque vectoring point of view, the driver's input simultaneously acts as a reference and as a disturbance. This paper presents a method to explicitly consider this aspect in the torque vectoring design by means of a feedforward compensation. At first, we identify the plant model highlighting the effects of the driver input steer on the system. Then, we design the compensation term with mathcal{H}_{infty} techniques based on the identified model. This solution is then refined using a textit{Data-Driven} approach based on a Bayesian optimization algorithm. Results show that the compensation term can reduce the tracking error by 40% and the control effort by 35%.

Keywords: Stochastic Systems, Multi-agent and Networked Systems
Abstract: This paper generalizes our earlier results published in the Proceedings of 2021 Modeling, Estimation and Control Conference. The paper also provides the result on semistability in mean square for switched nonlinear discrete-time systems. The theoretical results involve generalized sufficient conditions for (stochastic) semistability and semistability in mean square of discrete-time nonlinear dynamical systems under time-varying or random (arbitrary) switching by means of Fixed Point Theory. An advantage of these results is to overcome fundamental difficulties arising from existing methods such as Lyapunov and LaSalle methods. As an application of the theoretical results presented, a constrained distributed consensus problem over random multi-agent networks is considered. The random Krasnoselskii-Mann iteration is applied for the problem to derive a generalized asynchronous and totally asynchronous algorithm. The algorithm is able to converge even if the weighted matrix of the graph is periodic and irreducible under synchronous protocol. Finally, a numerical example is given in which there is a distribution dependency among communication graphs to demonstrate the results.

Keywords: Modeling and Validation
Abstract: We live in an increasingly electrified world. For stationary applications such as industry and manufacturing, this statement has been obvious since the start of the 20th century as steam and belt drives in factories gradually gave way to electric motors for machining, conveyor lines, and all manner of other industrial applications. Now we are seeing a dramatic rise in the electrification of mobility systems. The progress has been steady for several decades but it is really during the past several years that electrified mobility has seen a rapid growth at the level of individual consumer. Interestingly, this growth cuts across widely varying modes of mobility; from individual bicycles to on-highway vehicles to large ships and aircraft. This talk will detail some of the technical challenges related to Modeling, Control and Design. Of high relevance to systems and controls audiences is the interplay between modes of power distribution within electrified mobility systems. This includes the flow of power, or storage of energy, in the mechanical, chemical, electrical, and thermal domains. For example, power flow in the electrical domain can be constrained by component temperature limits in the thermal domain. Several examples of challenges will be raised along with some solutions for specific problems of Modeling, Control, and Design in electrified mobility. The presented solutions will be integrated such that the chosen modeling tools are specifically amenable to both the control and design challenges. Simulation and experimental results will be presented that demonstrate a superior overall mobility platform performance when a systems approach is taken.

Keywords: Manufacturing Systems, Modelling, Identification and Signal Processing, Mechatronic Systems
Abstract: Compared to conventional electronics fabrication techniques, printed electronics have the advantages of reducing complexity and cost, while increasing material compatibility and design flexibility. Electrohydrodynamic jet (E-jet) printing is a 3D printing process with high-resolution capabilities. While significant progress has been made on printing metal nanoparticles to fabricate micron-scale electrical components, a knowledge gap exists in our understanding of the key process strategies that enable e-jet printing at the submicron-scale. This work highlights the derivation of models for performance estimation and process parameter selection to ensure consistent e-jet printing at the submicron-scale. Using this framework, we demonstrate high-resolution conductive line patterns suitable for high-performance electronics applications.

Keywords: Modelling, Identification and Signal Processing, Mechatronic Systems, Manufacturing Systems
Abstract: We present a system identification method based on recursive least-squares (RLS) and coprime collaborative sensing, which can recover system dynamics from non-uniform temporal data. Focusing on systems with fast input sampling and slow output sampling, we use a polynomial transformation to reparameterize the system model and create an auxiliary model that can be identified from the non-uniform data. We show the identifiability of the auxiliary model using a Diophantine-equation approach, and show the stability using the hyperstability theory. Numerical examples demonstrate the successful system reconstruction and the ability to capture fast system response with limited temporal feedback.

Keywords: Power and Energy Systems, Estimation, Machine Learning in modeling, estimation, and control
Abstract: This paper addresses the state of charge estimation problem in lithium iron phosphate (LFP) battery cells. LFP cells are particularly challenging because their flat open circuit voltage (OCV) curve means OCV-based battery models are weakly observable. This means standard methods for SOC estimation don't easily converge to the true SOC. Additionally, in practice, estimates must be accurate in the face of biased noise on current input, as well as mean-zero noise on measurements. As such, we aim to create an estimator that is accurate when facing these types of noise. We accomplish this with a three-layer estimation technique that uses an adaptive Kalman filter, a Neural Network, and a Kalman Filter to estimate the state of charge. The proposed method achieves an SOC estimation with an RMSE of 2.248%, even in the presence of 0.2A current measurement bias and 5mA and 5mV random measurement noises. The proposed approach outperforms state-of-the-art methods like the extended Kalman filter.

Keywords: Modeling and Validation, Modelling, Identification and Signal Processing, Automotive Systems
Abstract: For off-road vehicles such as excavators and wheel loaders, a large portion of energy is consumed to overcome the soil resistant force in the digging process. For optimal control of the digging tool, a high-fidelity model of the soil-tool interaction force is important to reduce energy consumption. In this paper, an on-line soil resistant force prediction method is proposed. In this method, a hybrid model, which combines a physical model and a data-driven model, is used for the force prediction. In addition, the parameters of the hybrid model can be updated on-line based on real-time data. Comparisons with experimental data demonstrate that the proposed prediction method has an average error of around 12.7%.

Keywords: Automotive Systems
Abstract: One-pedal driving is a user-friendly feature, particularly in electric vehicles, allowing for simple, energy-efficient, comfortable, and safe driving. In this study, we propose a simple and effective method to generate a human-like braking speed profile in the presence of a red traffic light or a stop sign. This human-like braking profile can serve as a reference profile of a human driver at a given initial speed and distance-to-stop. The proposed method is developed based on four-phase braking and vehicle longitudinal dynamics. The method utilizes parameters that describe the braking behavior of individual drivers, such as the braking time for a given initial speed and distance-to-stop, the minimum acceleration, the minimum jerk, and the maximum jerk during the braking. These parameters are identified from real-world driving data. The results show that the proposed method successfully captures a human driver's braking behavior when coming to a full stop without a preceding vehicle. The median RMSEs of the speed profile and median distance errors from all vehicles are 0.0075 m and 0.789m/s on average.

Keywords: Path Planning and Motion Control, Intelligent Autonomous Vehicles, Uncertain Systems and Robust Control
Abstract: Generally, large discrepancies exist between predicted and realized energy consumption with energy-aware trajectory planning algorithms for off-road vehicles. Conservative planners typically pre-compensate for the expected discrepancy by demanding high confidence thresholds. Global path planners often ignore the substantial energy needed for turning on off-road deformable terrains, contributing to this mismatch. In this paper, we improve energy predictions by adding an additional energy cost for turning maneuvers in the global path planner and reformulate the high-confidence global planner's cost function to reduce conservatism. We couple the proposed global planner with a nominal local planner to show the robustness and improved performance compared to existing energy-aware motion planners for off-road vehicles.

Keywords: Transportation Systems, Estimation, Control Applications
Abstract: This paper develops a multi-stage estimation algorithm for vehicle trajectory tracking applications. Previously designed nonlinear observers for vehicle trajectory tracking lack either the ability to handle variable velocity or have a high sensitivity to sensor noise. To overcome these shortcomings, the original model of the non-ego vehicle is translated into three separate models for speed, orientation, and position. Three stable observers are designed for these models which are all shown to be stable and robust to uncertainties. The new estimation algorithm outperforms previous high-gain and LMI-based nonlinear observers. The developed observer is useful for collision prediction and avoidance applications.

Keywords: Optimal Control, Machine Learning in modeling, estimation, and control, Automotive Systems
Abstract: Connectivity and automation technology has the potential to greatly improve the coordination and efficiency of vehicles. For example, advanced mapping and vehicle to everything (V2X) connectivity offer the opportunity to leverage data from the road infrastructure and traffic to make real-time decisions on planning the vehicle velocity and powertrain control strategy that improve the energy efficiency and reduce emissions.

Most of the algorithms that optimize the vehicle speed planning utilize either deterministic methods such as dynamic programming or machine learning methods to solve the problem. These methods generally suffer from high computational and memory requirements, which makes online implementation challenging. In particular, dynamic programming requires significant memory storage to be able to save the optimized full route trajectory solution, while reinforcement learning methods operate online and therefore directly add to the online computation load.

A hierarchical multi-horizon optimization framework is proposed where a model predictive control (MPC) is integrated with via a neural network (NN) that approximates the terminal cost of the receding horizon optimization problem. The NN learns a full-route value function approximation using a training data set from several dynamic programming optimal results from full-route vehicle simulations with varying traffic light signal data, accounting for the variability in route information.

Simulations over five real-world routes show that the integrated NN-rollout algorithm outperforms stochastic methods based on reinforcement learning, while reducing the computational and memory requirements. The NN-rollout provides an exhaustive mapping of the terminal cost similar to a full-route dynamic programming by acting as a function approximation, but without the significant memory required to store a look-up table for the dynamic programming solution. Additionally, the NN approximation shows similar robustness to variation as the reinforcement learning method due to training on and learning data that includes variation in traffic light signal phase and timing, while only requiring a completely offline training process that is computationally more efficient

Keywords: Power and Energy Systems, Control Design, Estimation
Abstract: Model parameter estimation is an important subject in control engineering, including the field of battery management. Input excitation optimization has become an emerging topic lately to improve the accuracy of estimation. Traditional optimization approach suffers from a fundamental issue in parameter uncertainty, as the target parameters for estimation, often needed for computing the optimization objective and constraints, are intrinsically unknown. In this study, we introduce a non-dimensional approach to optimize excitations for estimating the health-related Li-ion battery electrochemical parameters. Guided by the Buckingham πtheorem, we derived a control-oriented non-dimensional battery model, excluding uncertain target parameters from the problem formulation. As a result, the optimization problem can be solved without any prior knowledge of target parameters. The applicable control input sequence can be recovered by rescaling the obtained non-dimensional sequence with the best available knowledge of the parameters. Furthermore, the proposed method reveals the fundamental impact of the unknown parameters on the solution of input optimization. In light of this finding, we propose two iterative excitation optimization strategies, which both significantly improve the robustness and reduce the complexity of the optimization problem. The proposed method can be generalized to solve other optimal control problems.

Keywords: Assistive and Rehabilitation Robotics, Human-Machine and Human-Robot Systems, Biomechanical Systems
Abstract: The paper explores the use of knee stiffness as a parameter in the design of wearable knee assistive devices for augmenting human postural balance. The knee moment-angle relationship is utilized to estimate the quasi-stiffness of the knee. The measurement methods are carefully chosen to be non-invasive without rigid joint attachment to allow observation of unimpeded quiet stance. The relationship between identified biomechanical parameters and computed stiffness estimates is analyzed, and the resulting estimates are employed in the controller design of a stiffness-based knee assistive device. The paper also investigates the biomechanical response of the human body to the modulation of applied stiffness in the presence of varied visual stimuli. This research is a crucial first step toward designing knee-based assistive devices to enhance human postural balance in destabilizing environments.

Keywords: Healthcare systems, Machine Learning in modeling, estimation, and control, Sensors and Actuators
Abstract: To improve monitoring of cerebral blood flow and arterial response to clinical interventions following acute traumatic brain injury, we have created a small sensor suite that can be mounted around the diameter of a typical intracranial catheter. This instrument is driven by the need to prevent subsequent ischemic injury after a traumatic brain injury, due to elevated intracranial pressure and compromised cerebral autoregulation. To minimize effects on clinician workflow, our sensors can be integrated into catheters that are currently being used in accordance with accepted standards of care. This paper describes the use of a combination of thin piezoelectric material, a photoplethysmogram, and a long short-term memory regression network to track cerebral blood flow fluctuations. The results show a correlation (R^2 = 0.76) between beat-to beat waveform features collected using our sensor suite and reference blood flow measurements by ultrasound imaging. This method we introduce may help give medical professionals timely data on patients' cerebral hemodynamic status and how well patients responded to clinical interventions.

Keywords: Path Planning and Motion Control, Robotics
Abstract: Jumping robots are interesting devices that offer several advantages in terms of navigation about cluttered environments. However, they present unique challenges with respect to their design and control. In this paper, we propose a design that uses four planar thrusters/propellers on a jumping robot. Such a system provides better maneuverability due to agility provided by the propellers to guide the motion. From energy consumption perspective, we use the gravity for free fall and a spring mechanism to execute the jumping motion without loss of much energy on impact. The primary contribution of this paper is developing a novel navigation and control method for such jumping robots to go to the desired goal. In this paper, we present a projectile motion planning approach for the control of the proposed system. We propose proportional (P) and proportional-derivative (PD) controllers that compute the launch velocity required for the jumping drone after impact with the ground to follow a projectile motion in each jump to reach the goal position. The jumping drone bounces after impact with the ground, and the drone is then actuated for a certain time till it attains the required launch velocity after which it is made to move freely under the influence of gravity. Such a system shows significant reduction of energy consumption 81.35% as compared to a normal drone navigating to the same goal location. Simulation results validate the effectiveness of the proposed approach.

Keywords: Modeling and Validation
Abstract: This paper presents the development and validation of a nonlinear delay differential equation (DDE) model for borehole propagation in the inclination plane. Most importantly, built upon a quasi-linear model, the nonlinear approach incorporates information pertaining to the floating stabilizers and bit tilt saturation by formulating a linear complementarity problem. As a result, the outputs of the nonlinear model were in a better agreement with the field data when compared with the quasi-linear model. The maximum modeling error of the nonlinear DDE is less than 1 degrees over a drilled depth of 600 feet.

Keywords: Automotive Systems, Motion and Vibration Control, Sensors and Actuators
Abstract: This paper explores the advantages of a combined handling-oriented stiffness/damping control for a multichamber suspension. It exploits the spring stiffness modulation capabilities to reduce the steady-state chassis angles of up to 49% in roll and 36% in pitch with respect to the passive benchmark and the damper regulation to prevent any overshoots or oscillations in the transient phase of harsh maneuvers, with an improvement of up to 44% in pitch rate and 29% in roll rate. Overall, the combined controller enhances the vehicle behavior while ensuring satisfactory comfort performance on single-events obstacles and irregular roads.

Keywords: Multi-agent and Networked Systems, Optimal Control, Modelling and Control of Biomedical Systems
Abstract: In this paper, the problem of identifying the topology of a linear dynamic network using data acquired from a mobile agent is considered. Network topology learning is important in various areas where not only the direction of the node connections, but also the time-dependent characteristics of these connections need to be identified. Mobile-sensing related issues such as the measurement cost, however, have not yet been accounted for in the existing work. The proposed approach combines the Wiener filter technique with gradient-based optimization to minimize the total estimation error in the output (i.e., the node response), and then, the cost in terms of the total measurement time. A simulation example of identifying a 6-node network is presented to illustrate the proposed approach. The simulation results show that for the network of inter-node dynamics of a wide range of characteristics (time constants), both the direction, existence, and dynamics of the inter-node connections can be accurately identified by using the proposed approach.

Keywords: Modelling and Control of Biomedical Systems, Healthcare systems, Biomechanical Systems
Abstract: This paper develops a scalable model of the dynamics of gas exchange during peritoneal oxygenation, motivated by the potential of such oxygenation to provide life support for patients with severe respiratory failure. The literature presents peritoneal oxygenation experiments for both large and small animals, including adult swine, rabbits, rats, and piglets. Results of these experiments suggest a potential discrepancy, with the benefits of peritoneal oxygenation possibly being stronger for smaller animals. We hypothesize that this size dependence is at least partially attributable to the effect of animal size on the ratio of peritoneal diffusion surface area to animal volume. The paper develops a scalable multi-compartment model of gas transport dynamics during peritoneal oxygenation. Simulating this model provides important insights regarding the potential impact of animal size on the viability of peritoneal oxygenation.

Keywords: Control Design, Nonlinear Control Systems, Manufacturing Systems
Abstract: Stabilizing a reference trajectory for a nonlinear system is a common, non-trivial task in control theory. An approach to solve this problem is to approximate the nonlinear system along the trajectory as an uncertain linear time-varying one, and to solve an optimization problem featuring Linear Matrix Inequality (LMI) constraints to derive a stabilizing, smooth, gain-scheduled control law. Such an approach is extended here by considering a set of reference trajectories instead of a single one, such that switching among them is permitted. These switching events are commonly encountered in industrial plants, such as energy generation systems, and are of high relevance in practice. The approach allows one to derive a gain-scheduled control law guaranteeing asymptotic stability also during the switching and accounting for the linearization errors. Simulation results on a chemical system highlight the effectiveness of the method.

Keywords: Control Design
Abstract: This paper focuses on the design of an individual blade pitch controller to reduce the platform oscillation of a floating offshore wind turbine with repositioning capacity. Such reduction is helpful for captured power regulation in wind farm control via turbine repositioning. In this study, a baseline 5MW wind turbine on a semi-submersible platform with long mooring lines is used for demonstration purposes. The model for the controller design is obtained as a state equation using the linearization functionality of the wind turbine simulator OpenFAST, activating only platform pitch and platform yaw as degrees of freedom. Based on the model, a linear quadratic regulator (LQR) is designed for platform vibration suppression. During each revolution, the controller prioritizes producing restoring moments to address platform pitch oscillation as the blades approach the vertical centerline of the rotor plane. Conversely, as the blades rotate close to the horizontal centerline of the rotor plane, the controller shifts its focus towards generating restoring moments to counteract platform yaw fluctuations. It is demonstrated in OpenFAST that the designed LQR-based individual blade pitch controller effectively reduces platform surge, roll, pitch, and yaw oscillation, resulting in improved power regulation during turbine repositioning.

Keywords: Human-Machine and Human-Robot Systems, Robotics, Sensors and Actuators
Abstract: Tactile skins made from textiles enhance robot-human interaction by localizing contact points and measuring contact forces. This paper presents a solution for rapidly fabricating, calibrating, and deploying these skins on industrial robot arms. The novel automated skin calibration procedure maps skin locations to robot geometry and calibrates contact force. Through experiments on a FANUC LR Mate 200id/7L industrial robot, we demonstrate that tactile skins made from textiles can be effectively used for human-robot interaction in industrial environments, and can provide unique opportunities in robot control and learning, making them a promising technology for enhancing robot perception and interaction.

Keywords: Stochastic Systems, Uncertain Systems and Robust Control, Automotive Systems
Abstract: In this paper, we focus on developing a multi-step uncertainty propagation method for systems with state- and control-dependent uncertainties. System uncertainty creates a mismatch between the actual system and its control oriented model. Often, these uncertainties are state- and control-dependent, such as modeling error. This uncertainty propagates over time and results in significant error over a given time horizon, which can disrupt operation of safety critical systems. Stochastic predictive control methods can ensure that the system stays within the safe region with a given probability, but requires prediction of the future state distributions of the system over the horizon. Predicting the future state distribution of systems with state- and control-dependent uncertainty is a difficult task. Existing methods only focus on modeling the current or one-step uncertainty, while the uncertainty propagation model over a horizon is generally over-approximate. Hence, we present a mutli-step Gaussian Process Regression method to learn the uncertainty propagation model for systems with state- and control-dependent uncertainties. We also perform a case study on vehicle lateral control problems, where we learn the vehicle's error propagation model during lane changes. Simulation results show the efficacy of our proposed method.

Keywords: Uncertain Systems and Robust Control, Adaptive and Learning Systems, Power and Energy Systems
Abstract: This article proposes an approach to address the current and the aggregated active power control challenge for large-scale inverter-based resources subjected to partially or completely loss of inverters and grid voltage variations. To address this problem, a distributed active power mechanism is proposed which generates desired inverters current. Then an adaptive mechanism distributes the voltage control input automatically between inverters in response to the partial or complete loss of inverters. An mathcal{L}_2-gain-based controller is designed for each inverter to track the desired current and rejects the grid voltage disturbance. Simulation results show a significant robust tracking for collective active power and current.

Keywords: Multi-agent and Networked Systems, Robotics
Abstract: This paper proposes a capture strategy for a team of pursuers to capture a fast evader with uncertain dynamics in a 3D reach avoid game. The strategy involves coordinating the agents motion to spread out and cut off all possible routes to the evader's target. These routes are obtained by estimating the evader's reachable set using mixed monotone reachable set theory. The reachable set is used to determine a capture surface, over which embedded guidance reference points are provided for the pursuers through 2D coverage. The capture strategy is demonstrated via simulation. Results suggest that capture performance improves with an increase in pursuer team size. Further, the strategy is able to outperform a pure-pursuit strategy when there is a sufficient number of pursuers to fully cover the obtained capture surface.

Keywords: Sensors and Actuators, Soft Robotics, Assistive and Rehabilitation Robotics
Abstract: Collecting gait data and providing haptic feedback are essential for the safety and efficiency of robot-based rehabilitation. However, readily available devices that can perform both are scarce. This work presents a novel method for mutual sensing and haptic feedback, through the development of an Inflatable Soft Haptic Sensor (ISHASE). The design, modeling and characterization of ISHASE are discussed. Four ISHASE are embedded in the insole of a shoe to measure ground reaction forces and provide haptic feedback. Four participants were recruited to evaluate the performance of ISHASE as a sensor and haptic device. Experimental results indicate that ISHASE can accurately estimate the user’s ground reaction forces while walking, with a maximum and a minimum accuracy of 91% and 85% respectively. Haptic feedback was delivered to four different locations under the foot and the users could identify the location with an average 92% accuracy. A case study, that exemplifies a rehabilitation scenario, is presented to demonstrate the ISHASE’s usefulness for mutual sensing and haptic feedback.

Keywords: Mechatronic Systems, Modelling, Identification and Signal Processing, Estimation
Abstract: This tutorial paper will focus on the author’s state space structures, the Biquad State Space (BSS) and the Bilinear State Space (BLSS). These two structures are have shown some remarkable advantages in the modeling and control of mechatronic systems, including numerical stability and model explainability. Furthermore, they have the remarkable property that the states of digital BSS and BLSS structures correspond to the states of the analog BSS and BLSS structures, at the outputs of each biquad or bilinear section. This gives the control engineer the ability to connect their digital model far more closely to the physical system, allowing digital “scope probes” to compare the same signals as one might get from the physical system.

Keywords: Control Design, Mechatronic Systems, Control Applications
Abstract: This tutorial paper will focus on revisiting issues of discretization. The accepted, theoretically correct method of discretizing a linear, time-invariant model in a feedback loop is with a hold equivalent, most commonly the zero-order hold (ZOH) equivalent. However, for the BSS and the BLSS to do their state-preserving magic, they must be discretized one block at a time. This flies in the current dogma that requires only a hold equivalent, typically a zero-order hold (ZOH) equivalent, can correctly discretize a feedback system. We will delve into the historical nature of this equivalent, as well as the conditions under which it represents the “exact” answer. We will also go through the handful of simple analytic examples used before one simply gives up and uses numerical means. We will also delve into what is lost in the name of mathematical exactitude. We finish with a comparison of the relative accuracy of discrete state-space forms for some illustrative examples, making the case that in many (and perhaps most) modern digital control systems, it may be worth giving up on mathematical exactitude in favor of having a very close system representation that can be understood and debugged.