Jeong, Jongryeol (Argonne National Laboratory), Kim, Namdoo (Argonne National Laboratory), Karbowski, Dominik (Argonne National Laboratory), Rousseau, Aymeric (Argonne National Laboratory)

Implementation of Model Predictive Control into Closed-Loop Micro-Traffic Simulation for Connected Automated Vehicle

Scheduled for presentation during the Regular Session "Autonomous Vehicle Control" (TuAT1), Tuesday, June 25, 2019, 12:10−12:30, Chambord

9th IFAC International Symposium on Advances in Automotive Control, June 23-27, 2019, Orléans, France

This information is tentative and subject to change. Compiled on April 26, 2024

Abstract

Connected automated vehicles can utilize and share driving information with other vehicles and infrastructure in real time. Applying plentiful vehicle-driving data enhances the safety, mobility and performance of connected automated vehicles. Adaptive cruise control, one of the connected automated vehicle achievements, optimizes the speed in a constrained environment to achieve the best fuel economy while maintaining the safety of the vehicle. Model predictive control is widely used to achieve both fuel savings and robustness. It is important to validate the performance, robustness and implementability of model predictive control under real-world conditions that reflect the features of connected automated vehicle driving. In this research, model predictive control using quadratic programming and dynamic programming was implemented in a newly developed micro-traffic simulation program. That program, RoadRunner, can simulate the various driving conditions such as intersection status, speed limit, and grade of the road. RoadRunner is based on a forward-looking vehicle simulation tool, Autonomie, which has various types of configurations and validated vehicle models based on vehicle dynamics that make the simulation more reliable. As a result, the controllers show 3-5 % fuel savings accompanying the applicability to the real world driving.