Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Learning a Stable, Safe, Distributed Feedback Controller for a Heterogeneous Platoon of Autonomous Vehicles

Michael H. Shaham, Taskin Padir

TL;DR

Experimental results demonstrate the practicality of the algorithm and the learned controller by comparing the performance of the neural network controller to linear feedback and distributed model predictive controllers.

Abstract

Platooning of autonomous vehicles has the potential to increase safety and fuel efficiency on highways. The goal of platooning is to have each vehicle drive at a specified speed (set by the leader) while maintaining a safe distance from its neighbors. Many prior works have analyzed various controllers for platooning, most commonly linear feedback and distributed model predictive controllers. In this work, we introduce an algorithm for learning a stable, safe, distributed controller for a heterogeneous platoon. Our algorithm relies on recent developments in learning neural network stability certificates. We train a controller for autonomous platooning in simulation and evaluate its performance on hardware with a platoon of four F1Tenth vehicles. We then perform further analysis in simulation with a platoon of 100 vehicles. Experimental results demonstrate the practicality of the algorithm and the learned controller by comparing the performance of the neural network controller to linear feedback and distributed model predictive controllers.

Learning a Stable, Safe, Distributed Feedback Controller for a Heterogeneous Platoon of Autonomous Vehicles

TL;DR

Experimental results demonstrate the practicality of the algorithm and the learned controller by comparing the performance of the neural network controller to linear feedback and distributed model predictive controllers.

Abstract

Platooning of autonomous vehicles has the potential to increase safety and fuel efficiency on highways. The goal of platooning is to have each vehicle drive at a specified speed (set by the leader) while maintaining a safe distance from its neighbors. Many prior works have analyzed various controllers for platooning, most commonly linear feedback and distributed model predictive controllers. In this work, we introduce an algorithm for learning a stable, safe, distributed controller for a heterogeneous platoon. Our algorithm relies on recent developments in learning neural network stability certificates. We train a controller for autonomous platooning in simulation and evaluate its performance on hardware with a platoon of four F1Tenth vehicles. We then perform further analysis in simulation with a platoon of 100 vehicles. Experimental results demonstrate the practicality of the algorithm and the learned controller by comparing the performance of the neural network controller to linear feedback and distributed model predictive controllers.
Paper Structure (18 sections, 16 equations, 6 figures, 1 algorithm)

This paper contains 18 sections, 16 equations, 6 figures, 1 algorithm.

Table of Contents

  1. Introduction
  2. Related Work
  3. Autonomous vehicle platooning
  4. Certifiable learning-based control
  5. Methodology
  6. Controller design
  7. Controller learning and verification
  8. Experiments
  9. Controller learning
  10. Baseline controllers
  11. Hardware experiments
  12. Simulation experiments
  13. Discussion
  14. Conclusion
  15. Appendix
  16. ...and 3 more sections

Figures (6)

  • Figure 1: The platoon of four F1Tenth vehicles in the test course.
  • Figure 2: Per episode results for the maximum Lyapunov condition violations given by \ref{['eq:lyap_pos_ver']} and \ref{['eq:lyap_dec_ver']} when running algorithm \ref{['alg:guided_alg']}. Results are shown only for $N = 2$ (convergence at episode 30) and $N = 3$ (convergence at episode 150); convergence occurred after 8 episodes for $N = 1$.
  • Figure 3: Platoon trajectory when using the learned neural network controller.
  • Figure 4: Average root-mean-square error for each follower vehicle over the ten trials. We calculate the root-mean-square error for each vehicle's position and velocity error over each trial, and then estimate the root-mean-square error's 95% confidence interval based on the results over the ten trials.
  • Figure 5: Simulated trajectories for a platoon of 100 vehicles when using the learned neural network controller. The lead vehicle tracks the velocity profile shown by the reference. The faded lines in the position plot depict the desired position of each vehicle.
  • ...and 1 more figures