| |
Last updated on December 11, 2024. This conference program is tentative and subject to change
Technical Program for Friday December 13, 2024
|
FrAT1 |
Room T1 |
Network Games, Learning, and Control for Sustainability and Climate Change |
Invited Session |
Co-Chair: Parasnis, Rohit | Massachusetts Institute of Technology |
Organizer: Parasnis, Rohit | Massachusetts Institute of Technology |
Organizer: Amin, Saurabh | Massachusetts Institute of Technology |
|
10:00-10:15, Paper FrAT1.1 | |
Equity and Equality in Repeated Weighted Congestion Games with Artificial Currency Incentives (I) |
|
Pedroso, Leonardo (Eindhoven University of Technology), Agazzi, Andrea (Università Di Pisa), Heemels, Maurice (Eindhoven University of Technology), Salazar, Mauro (Eindhoven University of Technology) |
Keywords: Public policies, Intelligent road transportation
Abstract: In scenarios where users access shared resources selfishly, the resulting societal costs often exceed those of a centrally coordinated optimal allocation. While monetary tolls have been used to steer user choices towards desired optima, they discriminate against lower-income users. Recent research has explored incentive schemes based on artificial currencies to achieve an allocation that is both system-optimal and fair. In this paper, we focus on the repeated weighted congestion game with two resources, where users contribute differently to congestion. After formally defining equity and equality, we devise weight-dependent and time-invariant optimal pricing policies that maximize each of these fairness metrics.
|
|
10:15-10:30, Paper FrAT1.2 | |
A Game-Theoretic, Market-Based Approach to Extract Flexibility from Distributed Energy Resources (I) |
|
Nair, Vineet (Massachusetts Institute of Technology), Annaswamy, Anuradha (Massachusetts Inst. of Tech) |
Keywords: Smart Grid and Demand Response, Smart cities, Smart infrastructure
Abstract: We propose a market designed using game theory to optimally utilize the flexibility of distributed energy resources (DERs) like solar, batteries, electric vehicles, and flexible loads. Market agents perform multiperiod optimization to determine their feasible flexibility limits for power injections while satisfying all constraints of their DERs. This is followed by a Stackelberg game between the market operator and agents. The market operator as the leader aims to regulate the aggregate power injection around a desired value by leveraging the flexibility of their agents, and computes optimal prices for both electricity and flexibility services. The agents follow by optimally bidding their desired flexible power injections in response to these prices. We show the existence and uniqueness of a Nash equilibrium among all the agents and a Stackelberg equilibrium between all agents and the operator. In addition to deriving analytical closed-form solutions, we provide simulation results for a small example system to illustrate our approach.
|
|
10:30-10:45, Paper FrAT1.3 | |
Operations of a Ride-Pooling Autonomous Mobility-On-Demand System in Mixed Traffic (I) |
|
Lucas, Clim (Eindhoven University of Technology), Paparella, Fabio (Eindhoven University of Technology), Cenedese, Carlo (ETH Zurich), Salazar, Mauro (Eindhoven University of Technology) |
Keywords: Urban mobility, Smart cities, Potential impact of automation and open problems
Abstract: We study optimal interventions and operational strategies for Ride-pooling Autonomous Mobility-on-Demand (R-AMoD) systems, whereby self-driving cars provide on-demand mobility to users that can either choose to pool together or alone, in the presence of private car users that selfishly optimize their routing choices. Specifically, we first frame the ride-pooling problem and operation of the R-AMoD system as a bi-level Stackelberg game, where the leader is the fleet operator controlling the R-AMoD system with the objective of minimizing the vehicle hours traveled by the fleet, and where the followers are private car owners that choose their routes by selfishly minimizing their own travel time. Second, we parse the problem and solve it with the BIG Hype algorithm: BIG Hype employs a combination of differentiation schemes to evaluate simultaneously the hypergradient and the lower-level equilibrium and its Jacobian to deliver a Stackelberg equilibrium with local optimality guarantees. Finally, we present a real-world case-study for Sioux Falls, USA, and compare our solution to a Nash equilibrium computed by iterating between the two problems until convergence.
|
|
10:45-11:00, Paper FrAT1.4 | |
Influencing Opinion Dynamics to Promote Sustainable Food Choices (I) |
|
Fontan, Angela (KTH Royal Institute of Technology), Eustachio Colombo, Patricia (London School of Hygiene and Tropical Medicine), Green, Rosemary (London School of Hygiene and Tropical Medicine), Johansson, Karl H. (KTH Royal Institute of Technology) |
Keywords: Public policies, Smart cities
Abstract: Substantial shifts in contemporary diets are needed to address the growing burden of chronic diseases and the accelerating climate crisis. To facilitate these changes, policy-makers must develop effective strategies to advertise and encourage healthy dietary choices among the population. In this work, to capture changes in dietary behavior we propose an opinion dynamics model where prejudiced agents can interact with their neighbors in a social network. Additionally, an external entity, such as a policy-maker or a government, can influence the agents through sequential campaigns towards a desired collective opinion. We investigate the impact of this exogenous influence under budget constraints, limiting, for example, the number of agents that can be influenced per campaign, and show how to optimize influence allocation. We conclude with an application of the proposed approach within the context of sustainability goals on food consumption set by the UK Climate Change Committee. Using baseline food intake patterns derived from survey data on the UK population, we evaluate the impact of campaigns promoting dietary shifts and mitigating environmental impacts.
|
|
11:00-11:15, Paper FrAT1.5 | |
Game-Theoretic Policy Intervention Design for Sustainable Forestry (I) |
|
Parasnis, Rohit (Massachusetts Institute of Technology), Amin, Saurabh (Massachusetts Institute of Technology) |
Keywords: Public policies
Abstract: We address the challenge of promoting sustainability in production forests, where a set of strategic entities is authorized to manage concession regions. These entities harvest commodities and sell them in the market. While concession owners (agents) often engage in sustainable activities that do not harm protected forest resources, they are also inclined to encroach upon protected areas to expand their agricultural operations, leading to unsustainable harvesting and production practices. Current sustainability certification programs aim to incentivize agents to reduce these unsustainable activities but fall short by ignoring the interdependence between sustainable and unsustainable activities, as well as the strategic interactions between neighboring agents. To overcome these shortcomings, we propose a coupled-activity network game model and design a budget-constrained policy that maximally reduces the aggregate level of unsustainable production in the network in the post-intervention equilibrium. Our analysis offers new insights into how pricing incentives can effectively suppress unsustainable production and sheds light on the influence of both intra-activity and cross-activity network effects on the agents' equilibrium production levels in both activities.
|
|
11:15-11:30, Paper FrAT1.6 | |
Strategic Control of Intersections for Efficient Traffic Routing without Tolls (I) |
|
Saltan, Yusuf (Bilkent University), Koşay, Arda (Bilkent University), Lin, Chung-Wei (National Taiwan University), Sayin, Muhammed Omer (Bilkent University) |
Keywords: Intelligent road transportation, Urban mobility, Smart cities
Abstract: Urban traffic congestion challenges sustainable transportation, causing economic losses (e.g., wasted fuel) and environmental damage. Two main factors contribute to this congestion: self-interested driver routing decisions, often misaligned with optimal traffic flow (as quantified by the Price of Anarchy), and intersections acting as bottlenecks with their limited resources. Prior research has addressed these issues separately, although an intersection’s (in)efficiency can also influence driver routing behavior. We propose strategic control of intersections based on priority-based scheduling as a non-monetary solution incentivizing socially optimal routing and ultimately reducing congestion. We quantify the performance in traffic simulations and evaluate its effectiveness in the well-studied Pigou’s example
|
|
11:30-11:45, Paper FrAT1.7 | |
Information Design for Congestion Minimization in Transportation Networks (I) |
|
Ambrogio, Alexia (Politecnico Di Torino), Cianfanelli, Leonardo (Politecnico Di Torino), Como, Giacomo (Politecnico Di Torino) |
Keywords: Intelligent road transportation, Urban mobility, Advanced control design-linear, non-linear, stochastic, large scale control systems
Abstract: We study an information design problem to reduce congestion in transportation networks. In presence of an uncertain network state, the central planner may sends private signals to the users, with the goal of steering the user equilibrium towards the system optimum flow. We consider private signals and provide sufficient conditions under which optimality may be achieved by information provision in networks with arbitrary number of parallel links and affine delay functions. Our results imply that optimality is more easily achieved in presence of large uncertainty of the random variables of the delay functions of the links of the network.
|
|
FrAT2 |
Room T2 |
Interactions of Pilot Handling & Autonomy for Next Generation Aircraft
Applications |
Invited Session |
Chair: Erturk, Sukru Akif | Turkish Aerospace |
Co-Chair: Demir, Mustafa | Turkish Aerospace |
Organizer: Erturk, Sukru Akif | Turkish Aerospace |
Organizer: Demir, Mustafa | Turkish Aerospace |
|
10:00-10:15, Paper FrAT2.1 | |
Adaptive Design Parameter Determination for Control Barrier Functions Using Reinforcement Learning (I) |
|
Memis, Sezer (Istanbul Technical University), Demir, Esra (Istanbul Technical University), Senel, Serkan (Istanbul Technical University), Demir, Mustafa (Turkish Aerospace), Koyuncu, Emre (Istanbul Technical University) |
Keywords: Automotive cooperated control (ADAS, etc), Advanced control design-linear, non-linear, stochastic, large scale control systems, Shared control
Abstract: This paper introduces an innovative method to improve control system performance through the adaptive determination of the design parameter (γ) in Control Barrier Functions (CBFs) using reinforcement learning (RL). Conventional approaches with a fixed γ parameter often fall short in dynamic environments. Our approach utilizes RL to dynamically adjust γ based on real-time feedback, allowing for more adaptable and efficient responses to varying conditions. Simulations in Adaptive Cruise Control (ACC) scenarios show that our adaptive γ selection significantly enhances the system's ability to maintain safety and performance. This method has promising implications for various safety-critical applications.
|
|
10:15-10:30, Paper FrAT2.2 | |
Modification of the Optimal Control Model of the Pilot and Its Application to the Selection of Inceptor Characteristics |
|
Grishina, Alyona (Moscow Aviation Institute), Efremov, Aleksandr (Moscow Aviation Institute), Efremov, Eugene (Moscow Aviation Institute) |
Keywords: Aircraft control, Human-Machine interaction in aircraft, Process plant operation
Abstract: The paper presents a modification of the optimal model of the characteristics of the pilot's control actions, which allows obtaining mathematical modeling results close to the results of experimental studies. This modification allows taking into account the characteristics and types of control inceptors, as well as additional proprioceptive feedback introduced by the pilot. Using the modified model, the optimization of control stick parameters providing the minimum variance of the error signal was performed, the results of which correspond to the results of experimental studies.
|
|
10:30-10:45, Paper FrAT2.3 | |
Robust Motion Planning for a Differentially Flat Fixed-Wing Aircraft (I) |
|
Aldabbas, Samer Raed (Turkish Aerospace Industries, Istanbul Technical University), Abu-Khalaf, Murad (Turkish Aerospace), Koyuncu, Emre (Istanbul Technical University) |
Keywords: Spacecraft control
Abstract: This paper investigates robustness and sensitivity in autonomous flight trajectory planning, focusing on model uncertainties during frame transitions. We analyze the role of differential flatness in planning open-loop state-control trajectories and examine how uncertainties impact robustness when mapped from the wind frame to the body frame. Using the given coordinated flight model, we assess mapping errors under mass uncertainty, reflecting these uncertainties back to the wind frame for analysis. Numerical simulations of a barrel roll maneuver illustrate our findings, contributing to the development of resilient trajectory planning frameworks for autonomous flight under variable conditions.
|
|
10:45-11:00, Paper FrAT2.4 | |
Exploring the Design Philosophy and Challenges in Cyber -Physical-Human Systems a Focus on Autonomous Terrain Following (I) |
|
Yurtsever, Mehmet Gorkem (Turkish Aerospace Industries), Mor, Zeynep Seda (Turkish Aerospace Industries), Tasdemir, Ataberk (Turkish Aerospace Industries), Buyukekiz, Kadir Bulathan (Turkish Aerospace Industries), Bas, Emre (Turkish Aerospace Industries), Erkek, Bedirhan Burak (Turkish Aerospace Industries), Topuk, Seymanur (Turkish Aerospace Industries) |
Keywords: Aircraft control, Human-Machine interaction in aircraft
Abstract: This paper explores Terrain Following (TF) design in autonomous flight systems, focusing on terrain smoothing, flight path adjustments, and pilot comfort and safety. Key elements include processing terrain data, generating guidance commands, and maintaining pilot trust. Various TF flight modes are analyzed to balance smooth flying with detectability. The study offers insights into enhancing safety and efficiency in aviation through improved autonomous systems.
|
|
11:00-11:15, Paper FrAT2.5 | |
Enhanced Flight Envelope Protection: A Novel Reinforcement Learning Approach (I) |
|
Catak, Akin (Istanbul Technical University), Altunkaya, Ege Cagri (Istanbul Technical University), Demir, Mustafa (Turkish Aerospace), Koyuncu, Emre (Istanbul Technical University), Özkol, İbrahİm (Aerospace Research Center, İtÜ) |
Keywords: Aircraft control, Human-Machine interaction in aircraft, Shared control, Advanced control design-linear, non-linear, stochastic, large scale control systems
Abstract: This paper introduces a flight envelope protection algorithm on a longitudinal axis that leverages reinforcement learning (RL). By considering limits on variables such as angle of attack, load factor, and pitch rate, the algorithm counteracts excessive pilot or control commands with restoring actions. Unlike traditional methods requiring manual tuning, RL facilitates the approximation of complex functions within the trained model, streamlining the design process. This study demonstrates the promising results of RL in enhancing flight envelope protection, offering a novel and easy-to-scale method for safety-ensured flight.
|
|
11:15-11:30, Paper FrAT2.6 | |
Development of Means for Suppressing Negative Aircraft-Pilot Coupling Effects in Manual Control |
|
Efremov, Aleksandr (Moscow Aviation Institute), Scherbakov, Aleksandr (Moscow Aviation Institute), Efremov, Eugene (Moscow Aviation Institute) |
|
|
11:30-11:45, Paper FrAT2.7 | |
Active Sidestick Control Integration for Enhanced Aircraft Flight Envelope Protection (I) |
|
Altunkaya, Ege Cagri (Istanbul Technical University), Erol, Fatih (AIGNC Research Group, Aerospace Research Center, Istanbul Techni), Catak, Akin (Istanbul Technical University), Mert, Volkan (Turkish Aerospace), Capone, Pierluigi (Zurich University of Applied Sciences), Erturk, Sukru Akif (Turkish Aerospace), Koyuncu, Emre (Istanbul Technical University) |
Keywords: Aircraft control, Human-Machine interaction in aircraft, Decision-support for human operators, Shared control
Abstract: The design of Envelope and Pilot-Induced Oscillation (PIO) Protection Features, and Failure Cases detection and prevention using Active Control Sidestick (ACS) is a challenging task. While helping the pilot to respect the envelope limitations also in failure scenarios and, therefore, increasing mission effectiveness, these features may have a significant impact on the aircraft's agility. ACS characteristics are investigated in an integrated environment. A set of effective and flexible control laws based on Incremental Nonlinear Dynamic Inversion have been developed in a state-of-the-art aircraft simulation model and coupled with a two-ways communication with the selected ACS. The model can run in real-time in a fixed-based simulator composed of representative cockpit and out-of-the-window.
|
|
FrBT1 |
Room T1 |
Autonomous Vehicles |
Regular Session |
Chair: Vardar, Yasemin | TU Delft |
Co-Chair: Ozay, Necmiye | University of Michigan |
|
13:30-13:45, Paper FrBT1.1 | |
Human-Inspired Learning for Car Following Models |
|
Al Habboush, Seymanur (Bilkent University), Yildiz, Yildiray (Bilkent University), Annaswamy, Anuradha (Massachusetts Inst. of Tech) |
Keywords: Intelligent road transportation
Abstract: In this paper, we propose a human-inspired learning mechanism in the context of car following models. We use a memory structure to gather information from other drivers and make inferences about their driving styles. Then, this information is used to determine the ideal driving strategy. Subsequently, the learning process between the current and the ideal driving strategies is modeled with the help of adaptive control techniques. Finally, we incorporate the proposed learning mechanism into a multi-type car following model that we introduce. The performance of the proposed method is investigated using the NGSIM traffic data set.
|
|
13:45-14:00, Paper FrBT1.2 | |
ADAS Evolution: LSTM-Driven Autonomous Adaptation to Individual Driver Preferences |
|
Wang, Yikai (Univ. Polytechnique Hauts-De-France, CNRS, UMR 8201 – LAMIH - La), Ota, Naoto (University of Tsukuba), Debernard, Serge (UPHF - Universite Polytechnique Hauts De France), Popieul, Jean-Christophe (University of Valenciennes/LAMIH) |
Keywords: Automotive cooperated control (ADAS, etc), Intelligent road transportation, Decision-support for human operators
Abstract: The ongoing advancement of advanced driver assistance systems (ADAS) has led to notable improvements in vehicle safety and efficiency within the automotive industry (Kala, 2016). Nevertheless, these systems also present new challenges and demands for developers, especially concerning driver trust and acceptance, which are vital for the safety and efficiency of collaborative human-machine driving. In response to these problems, the CoCoVéIA project proposes an innovative approach in which the driver assistance system (ADAS) integrates continuous learning of individual driving preferences, to boost driver confidence and acceptance. At the core of this method is the use of autonomous learning algorithms, particularly long short-term memory (LSTM) networks, which are adept at learning and adjusting to different driving preferences in real-time. This paper confirms the efficacy of LSTM algorithms in precisely learning and predicting driving preferences through experimental data, showcasing their ability to accommodate a variety of driver behaviors and preferences.
|
|
14:00-14:15, Paper FrBT1.3 | |
Analysis of Human Steering Behavior Differences in Human-In-Control and Autonomy-In-Control Driving |
|
Mai, Rene (Rensselaer Polytechnic Institute), Julius, Agung (Rensselaer Polytechnic Institute), Mishra, Sandipan (Rensselaer Polytechnic Institute) |
Keywords: Automotive cooperated control (ADAS, etc), Semi-autonomous and mixed-initiative systems, Shared control
Abstract: Steering models (such as the generalized two-point model) predict human steering behavior well when the human is in direct control of a vehicle. In vehicles under autonomous control, human control inputs are not used; rather, an autonomous controller applies steering and acceleration commands to the vehicle. For example, human steering input may be used for state estimation rather than direct control. We show that human steering behavior changes when the human no longer directly controls the vehicle and the two are instead working in a shared autonomy paradigm. Thus, when a vehicle is not under direct human control, steering models like the generalized two-point model do not predict human steering behavior. We also show that the error between predicted human steering behavior and actual human steering behavior reflects a fundamental difference when the human directly controls the vehicle compared to when the vehicle is autonomously controlled. Moreover, we show that a single distribution describes the error between predicted human steering behavior and actual human steering behavior when the human's steering inputs are used for state estimation and the vehicle is autonomously controlled, indicating there may be a underlying model for human steering behavior under this type of shared autonomous control. Future work includes determining this shared autonomous human steering model and demonstrating its performance.
|
|
14:15-14:30, Paper FrBT1.4 | |
Development of a Novel Adaptive Driver Assistance Based on Real-Time Cornering Parameter Identification |
|
Keleş, Ahmet (Ihsan Dogramaci Bilkent University), Dağ, Doğa (Ihsan Dogramaci Bilkent University), Cakmakci, Melih (Bilkent University) |
Keywords: Automotive cooperated control (ADAS, etc)
Abstract: In this paper, a novel adaptive driver assistance system was presented based on real-time cornering parameter identification using deep learning. Modern vehicle safety systems, like ESP (electronic stability program) and ADAS (advanced driving assistance systems), operate under specific conditions, requiring an adaptable system to perform as intended in all conditions. Our approach addresses this gap by continuously evaluating and adapting the vehicle’s response to real-time traction conditions. We introduce a cornering coefficient identification method using deep learning for instantaneous parameter estimation and combine it with an advanced driver assistance system. This system adapts driver inputs to current road conditions while maintaining stability and desired vehicle dynamics. The paper details the mathematical modeling of a four-wheel-drive, four-wheel-steering vehicle, the development of the parameter identification method, and the implementation of the adaptive driver assistance system. Experimental results demonstrate the system’s capability to enhance vehicle handling and safety across various driving conditions.
|
|
14:30-14:45, Paper FrBT1.5 | |
FITS: Ensuring Safe and Effective Touchscreen Use in Moving Vehicles |
|
Pool, Daan Marinus (Delft University of Technology, Faculty of Aerospace Engineering), Vardar, Yasemin (TU Delft) |
Keywords: Aircraft control, Human-Machine interaction in aircraft
Abstract: Touch interfaces are replacing physical buttons, dials, and switches in the new generation of cars, aircraft, and vessels. However, vehicle vibrations and accelerations perturb finger movements and cause erroneous touchscreen inputs by users. Furthermore, unlike physical buttons, touchscreens cannot be operated by touch alone and always require users’ visual focus. Hence, despite their numerous benefits, touchscreens are not inherently suited for use in vehicles, which results in an increased risk of accidents. In a recently awarded research project, titled "Right Touch Right Time: Future In-vehicle Touchscreens (FITS)", we aim to address these problems by developing novel in-vehicle touchscreens that actively predict and correct perturbed finger movements and simulate physical touch interactions with artificial tactile feedback.
|
|
14:45-15:00, Paper FrBT1.6 | |
4C: Custom-And-Correct-By-Construction Controller Synthesis Using Multi-Modal Human Feedback |
|
Karagulle, Ruya (University of Michigan), Valdez Calderon, Marco Antonio (University of Michigan), Ozay, Necmiye (University of Michigan) |
Keywords: Potential impact of automation and open problems
Abstract: Autonomous vehicle technology has a critical role in enhancing road safety by reducing human errors. The adoption of such technologies is shown to be dependent on user satisfaction, which is influenced by diverse comfort and performance preferences. This paper addresses the challenge of generating custom and naturalistic autonomous vehicle behaviors to meet individual user expectations while ensuring safety. To achieve this goal, we incorporate multi-modal human feedback with pairwise preferences and driving demonstrations into a control synthesis framework. We write safety specifications and express preferences using Weighted Signal Temporal Logic (WSTL), and we employ the pytelo~toolbox for controller synthesis. Our approach is validated through synthetic experiments and a pilot, small-sized human subject study, demonstrating the effectiveness of integrating multi-modal human feedback for customizing autonomous vehicle behaviors.
|
|
FrBT2 |
Room T2 |
Assistive Devices and Methods |
Regular Session |
Co-Chair: Kille, Sean | Karlsruhe Institute of Technology (KIT) |
|
13:30-13:45, Paper FrBT2.1 | |
Human Decisions versus Multi-Agent Reinforcement Learning in a Pursuit-Evader Game |
|
Adams, Takuma A (The University of Melbourne), Cullen, Andrew C. (University of Melbourne), Alpcan, Tansu (The University of Melbourne) |
Keywords: Decision-support for human operators, Potential impact of automation and open problems, Advanced control design-linear, non-linear, stochastic, large scale control systems
Abstract: The modelling and analysis of nonlinear cyber-physical systems is integral to applications ranging from social networks to defence strategy. However, conventional linear-quadratic methods inadequately capture the complex nonlinear behaviour of human decision-makers in these systems. Utilising game theory and multi-agent reinforcement learning (MARL), we explore dynamic interactions between humans and machines within a novel cyber-physical environment, using a swarmalator to represent strategic humans in a pursuit-evader game. Preliminary results indicate that MARL inadequately captures the nonlinear behaviour of human decision-making, with pursuers and evaders exhibiting human behaviours achieving the highest utilities.
|
|
13:45-14:00, Paper FrBT2.2 | |
Application of Decision-Making Model Involving Irrationality to Service Robot |
|
Munakata, Shinon (Tokyo University of Science), Hashimoto, Takuya (Tokyo University of Science), Kitano, Keisuke (Tokyo University of Science), Ichikawa, Hiroko (Tokyo University of Science), Kogiso, Kiminao (University of Electro-Communications) |
Keywords: Potential impact of automation and open problems, Assistive robotics, Comfort control in homes
Abstract: The irrational decision-making elicited by emotion is one aspect of humanity. The purpose of this study is to confirm the irrational behavior of robot affect human impression on the robot through as an actual interaction between human and robot. First, we made a conversation scenario between human and robot assuming a customer service situation in fast-food restaurant where an android robot works as a waitstaff. Then, a decision-making model was designed for that setting. The experimental result showed that the irrational (emotional) behaviors caused by the decision-making model influenced human impression of robots in terms of anthropomorphism and likability.
|
|
14:00-14:15, Paper FrBT2.3 | |
Listening Effort Measurement in Unilateral Cochlear Implant Users: A Pupillometry Analysis |
|
Ayar, Ceylan (Hacettepe University), Senli, Fahrettin Deniz (Ankara Yildirim Beyazit University), Cicek Cinar, Betul (Hacettepe University), Acarturk, Cengiz (Jagiellonian University) |
Keywords: Biomedical implants, Neurostimulation
Abstract: Listening effort refers to the cognitive resources allocated to process auditory stimuli. While cochlear implants (CIs) restore hearing for individuals with profound loss, they often generate degraded speech signals due to poor spectral resolution. This study compared listening effort between 14 unilateral CI users and 14 normal-hearing participants using pupillometry and response time measurements. Although advancements in technology, CIs continue to present difficulties in complicated listening situations, requiring additional efforts from users. Our findings show that CI users expend more effort than normal-hearing individuals, even when their speech intelligibility scores are comparable.
|
|
14:15-14:30, Paper FrBT2.4 | |
Validation of Human-Variability Respecting Optimal Control: A Preparational Study |
|
Kille, Sean (Karlsruhe Institute of Technology (KIT)), Sobeloff, Leon (Karlsruhe Institute of Technology (KIT)), Fritz, Sebastian Andreas (Karlsruhe Institute of Technology (KIT)), Varga, Balint (Karlsruhe Institute of Technology (KIT), Campus South), Hohmann, Soeren (KIT) |
Keywords: Shared control, Assistive robotics, Semi-autonomous and mixed-initiative systems
Abstract: In order to design a performant and especially well perceived automation to assist the human in physical Human-Machine Interaction, an understanding of the human natural behavior is crucial. Model-based control design is therefore a favorable solution, already being used in various applications. However, most approaches model the human to be deterministic in it's behavior, not aligning to the state-of-the-art models presented by neuroscience, which more accurately describe the human to be under the influence of noise processes. This paper presents a study that examines 13 participants performing point-to-point movements, followed by the identification of the underlaying cost as well as additive and multiplicative noise process parameters of a human stochastic model. The gained results provide a basis for future automations that explicitly incorporate human variability in their control design.
|
|
FrCT1 |
Room T1 |
Advanced Control for CPHS II |
Regular Session |
Chair: Namerikawa, Toru | Keio University |
Co-Chair: Qu, Zhihua | University of Central Florida |
|
15:30-15:45, Paper FrCT1.1 | |
Human-In-The-Loop Adaptive Control Allocation for Uncertain Systems with Unknown Effector Degradation |
|
Sarioglu, N. Eren (Embry-Riddle Aeronautical University), Vernyi, Kyle (Embry-Riddle Aeronautical University), Dogan, Kadriye Merve (Embry-Riddle Aeronautical University) |
Keywords: Aircraft control, Human-Machine interaction in aircraft
Abstract: Adaptive control allocation methods are trending solutions for over-actuated uncertain systems. Thus, this work proposes an adaptive control allocation solution for compensating the effects of unknown effector degradation and model uncertainties in a humanin-the-loop closed-loop control system. Here, the human operator is modeled and used as a general transfer function. In addition, a low-frequency learning method is added to deal with the high-frequency control response to obtain a better transient response. The stability analysis is provided with Lyapunov theory which results in asymptotic tracking. Simulation results are provided on a nonlinear uncertain hexacopter attitude dynamics, where the effector degradation and the center of gravity (CoG) location are considered as unknown to show the efficacy of the adaptive method, where human output with three different transfer functions is used to produce a pitch rate command.
|
|
15:45-16:00, Paper FrCT1.2 | |
Human Supervision of High-Gain Observer-Based Control of Uncertain Nonlinear Multi-Agent Systems |
|
McCorkle, Patrick (Virginia Tech), Gautam, Kiran (Virginia Tech), Ramiandrisoa, Rojo (Virginia Tech), Boker, Almuatazbellah (Virginia Tech) |
Keywords: Shared control, Decision-support for human operators, Aircraft control, Human-Machine interaction in aircraft
Abstract: As tomorrow's machines become more general and autonomous, building resiliency into these systems is of critical importance. High-gain observers aggressively respond to errors and guarantee semi-global stability. Supervisory shared control architectures demonstrate the utility of a human-on-the-loop approach to enhance controllers via configurable parameters. Here, we investigate how a cascade high-gain observer-based controller designed to achieve leader-follower consensus can be modified to allow for human intervention, with the specific objective of maintaining performance goals while minimizing control saturation and aggressiveness. We perform envelope-straining experiments that suggest an adjustment of the high-gain parameter before the saturation limits is a better strategy to minimize aggression and thus energy cost.
|
|
16:00-16:15, Paper FrCT1.3 | |
Reconfiguration-Based Fault Tolerant Control Algorithm for LPV Descriptor Systems |
|
Deng, Hao (University Paris-Saclay), Stoica, Cristina (CentraleSupélec/Laboratoire De Signaux Et Systèmes), Chadli, Mohammed (University Paris-Saclay, Univ Evry, IBISC, Evry France) |
Keywords: Advanced control design-linear, non-linear, stochastic, large scale control systems
Abstract: This paper presents a fault tolerant control algorithm for linear parameter varying descriptor systems using a reconfiguration-based technique. The proposed reconfiguration block comprises a virtual actuator and a virtual sensor. The virtual actuator and virtual sensor gains are calculated by solving a linear matrix inequality problem with respect to the input-to-state stability criterion. The main advantage of the proposed method is that it is able to achieve fault tolerance without modifying the original nominal controller. The effectiveness of the the proposed approach is illustrated via a simulation example.
|
|
16:15-16:30, Paper FrCT1.4 | |
Resilient Cooperative Control of Passivity-Short Systems against Man-In-The-Middle Attacks |
|
Marasini, Ganesh (University of Central Florida), Qu, Zhihua (University of Central Florida), Mejia-Ruiz, Gabriel E (University of Central Florida) |
Keywords: Advanced control design-linear, non-linear, stochastic, large scale control systems, Smart Grid and Demand Response
Abstract: In this paper, the cooperative control problem is considered for heterogeneous subsystems of passivity short dynamics and in the presence of Man-in-the-Middle (MitM) attacks. In such a cooperative system, MitM attacks are false data injection (FDI) attacks through local communication channels or control mode alteration (CMA) attacks through tempering at distributed subsystems. A distributed and resilient cooperative control framework is proposed, and it consists of two hidden network layers, one being competitive interaction network, and the other being distributed observers. The proposed solution ensures robust consensus under attacks and also identifies the nature and locations of MitM attacks anywhere in the overall networked system. Furthermore, the proposed framework enables all the subsystems to gain situational awareness regarding potential attacks. These results are illustrated by a microgrid control application.
|
|
16:30-16:45, Paper FrCT1.5 | |
A Symmetric Divergence Measure in the Statistical Linearization |
|
Chernyshov, Kirill (V.A. Trapeznikov Institute of Control Sciences) |
Keywords: Advanced control design-linear, non-linear, stochastic, large scale control systems
Abstract: The problem of identifying the input-output mapping of multidimensional discrete-time stochastic systems using an information-theoretic criterion is considered. As an information-theoretic measure of dependence, which is the basis for constructing the corresponding criterion for statistical linearization, a measure based on the symmetric Kullback-Leibler divergence is proposed, in contrast to the usual Kullback-Leibler divergence, which, as is well known, is not symmetric. Based on such symmetric divergence, a corresponding measure of stochastic dependence is constructed, and the criterion of statistical linearization is the condition of pairwise coincidence of this measure of dependence, calculated for each component of the vector of the output variable of the system and each component of the vector of the input variable of the system, on the one hand, and the corresponding components of the output vector variable of the linearized model and the vector of the input variable of the system. This approach allows us to construct explicit analytical expressions for the weighting coefficients of the linearized model, which, at the same time, has the property of complete equivalence to the nonlinear system under study in the sense of coincidence of the corresponding information-theoretic characteristics.
|
|
16:45-17:00, Paper FrCT1.6 | |
Security Measure Implementation for Distributed State Estimation |
|
Shinohara, Takumi (Keio University), Namerikawa, Toru (Keio University) |
Keywords: Advanced control design-linear, non-linear, stochastic, large scale control systems, Potential impact of automation and open problems
Abstract: Toward the realization of secure and resilient cyber-physical human systems (CPHS), this paper deals with the distributed secure state estimation problem considering security measure implementation for the systems. We present that the implementation enhances the system resilience, which means that the distributed state estimation can be achieved even if more networks are compromised. We also show that the implementation is coNP-hard, but, on the positive side, the implementation can be solved in polynomial time when the system satisfies a certain condition. This paper further provides a distributed secure state estimator exploiting the security implementation.
|
|
FrCT2 |
Room T2 |
Games and Learning for CPHS |
Regular Session |
Chair: Basar, Tamer | Univ. of Illinois Urbana-Champaign |
Co-Chair: Li, Sarah H.Q. | University of Washington |
|
15:30-15:45, Paper FrCT2.1 | |
On the Impact of Bounded Rationality in Strategic Data Gathering |
|
Anand, Anju (Binghamton University), Akyol, Emrah (Binghamton University) |
Keywords: Public policies
Abstract: This paper is concerned with the problem of debiasing survey data gathered from strategic and boundedly rational agents with heterogeneous objectives and partial information. Particularly, we explore a setting where there are K different types of survey responders with varying levels of available information, degree of strategic behavior, and cognitive hierarchy: i) a non-strategic agent with an honest response, ii) a k-th level kin [1:K-1] strategic agent that believes the population is Poisson distributed over the lower cognitive types. We model each of these scenarios as a strategic classification of a 2-dimensional source (possibly correlated source and bias components) with quadratic distortion measures and provide a design method. We analyze the numerical results obtained via the proposed method whose implementation is available for research purposes at https://github.com/strategic-quantization/bounded-rationality.
|
|
15:45-16:00, Paper FrCT2.2 | |
Remote Estimation Games with Random Walk Processes: Stackelberg Equilibrium |
|
Dökme, Atahan (Bogazici University), Velicheti, Raj Kiriti (UIUC), Bastopcu, Melih (Bilkent University), Basar, Tamer (Univ. of Illinois Urbana-Champaign) |
Keywords: Advanced control design-linear, non-linear, stochastic, large scale control systems, Remote operation of robotic teams
Abstract: Remote estimation is a crucial element of real time monitoring of a stochastic process. While most of the existing works have concentrated on obtaining optimal sampling strategies, motivated by malicious attacks on cyber-physical systems, we model sensing under surveillance as a game between an attacker and a defender. This introduces strategic elements to conventional remote estimation problems. Additionally, inspired by increasing detection capabilities, we model an element of information leakage for each player. Parameterizing the game in terms of uncertainty on each side, information leakage, and cost of sampling, we consider the Stackelberg Equilibrium (SE) concept where one of the players acts as the leader and the other one as the follower. By focusing our attention on stationary probabilistic sampling policies, we characterize the SE of this game and provide simulations to show the efficacy of our results.
|
|
16:00-16:15, Paper FrCT2.3 | |
The Cyber-Placebo: Intelligent CPHS with Fake-AI? |
|
Gabrecht, Marco (Technical University of Applied Sciences Lübeck: Lübeck, DE), Janneck, Monique (Technische Hochschule Luebeck), Hellbrueck, Horst (TH Luebeck), Matthies, Denys J.C. (Technical University of Applied Sciences Lübeck) |
Keywords: Potential impact of automation and open problems, Ethics, Flexible manufacturing
Abstract: The placebo effect, a well-known phenomenon in medical and psychological research, is now being explored in the realm of Cyber-Physical & Human Systems (CPHS) with the introduction of the concept of a Cyber-Placebo. This approach involves simulating the presence and functionality of an artificial intelligence (AI) component within a CPHS, while in reality, a human operator, the wizard, controls the system’s responses and behaviors. The purpose of the Cyber-Placebo is three-fold: to evaluate user interaction and satisfaction with the system, investigate perceptions of moral responsibility and accountability of AI agents, and assess user responses to the risks and benefits of AI technology. This concept raises intriguing questions about measuring, manipulating, and utilizing a Cyber-Placebo, as well as connected ethical considerations. By bridging the gap between human perception and technological advancement, Cyber-Placebos may offer a unique opportunity to explore the complexities of human-machine interactions and pave the way for future developments in AI-driven CPHS.
|
|
16:15-16:30, Paper FrCT2.4 | |
Markov Potential Game with Reach-Avoid Objectives |
|
Li, Sarah H.Q. (University of Washington), P. Vinod, Abraham (Mitsubishi Electric Research Laboratories), Lygeros, John (ETH Zurich) |
Keywords: Next-generation air traffic management, Urban mobility, Advanced control design-linear, non-linear, stochastic, large scale control systems
Abstract: Combining potential game theory and stochastic reach-avoid problems, we introduce a Markov game with reach-avoid objectives with the goal of modeling multi-modal and mixed-autonomy mobility systems. Specifically, we propose a Markov game over the finite state-action space that extends the static reach-avoid objective to avoid other players’ trajectories. We motivate the existence of a potential function, propose an iterative best response extension of the single-player multiplicative dynamic programming, and empirically verify our algorithm on a two player reach-avoid Markov game.
|
|
16:30-16:45, Paper FrCT2.5 | |
Learning-Based Characterization of Noise Statistics for SoC Estimation Via Kalman Filtering |
|
Erdem, Ahmet Can (Turkish Aerospace), Tekİn, BariŞ (Istanbul Technical University), Mert, Volkan (Turkish Aerospace), Kocabas, Derya Ahmet (Istanbul Technical University), Altun, Tuncay (Yozgat Bozok University) |
Keywords: Smart Grid and Demand Response, Smart cities, Potential impact of automation and open problems
Abstract: In the field of Battery Management Systems (BMS), the State of Charge (SoC) is a crucial metric that represents the available energy capacity and directly affects the operational strategy. Accurately determining SoC is inherently complex due to the electrochemical characteristics that exhibit non-linear responses under various operational conditions. BMS operates model-based on an Extended Kalman Filter (EKF). In contrast, the employed Deep Deterministic Policy Gradient (DDPG) algorithm represents a model-free reinforcement learning methodology. This study aims to investigate the learning-based characterization of the process and measurement noise statistics for state of charge estimation via Kalman filtering. The method iteratively updates a value-function based on a reward mechanism, facilitating the selection of actions that minimize estimation error without a model of the environment. The model proposed in this study enhances EKF, a method known for its robustness in tracking SoC, by incorporating a RL paradigm. This paradigm is tailored to optimize parameter estimation despite sparse datasets. The adaptive mechanism is governed by a reward function that is based on minimizing SoC estimation error. This represents a judicious calibration between model-based and data-driven estimation techniques. The integration of DDPG improves our model's adaptability to SoC dynamics, promising enhanced estimation accuracy and improved reliability and efficiency in BMS across diverse applications.
|
| |