Adaptive MPC for Autonomous Driving - Evaluation on Fleet of Heavy-Duty Vehicles
IEEE Transactions on Intelligent Vehicles(2024)
摘要
This work conducts a systematic experimental evaluation of the state-of-the-art Reference Aware Model Predictive Controller (RA-MPC) for autonomous vehicles. The RA-MPC is a path-tracking controller, that maximizes tracking accuracy and comfort. The controller uses a kinematic vehicle model with a nonlinear curvature response table that adapts the steering response online to the vehicle and operating conditions. The adaptiveness and robustness of the controller are analyzed by evaluating the performance on a highway truck, loaded and empty mining trucks, and a city bus. Moreover, highway-like and city-like scenarios are performed using the exact same implementation and parameter settings for all vehicles. The controller and model adaption achieved a very good path tracking performance in all experiments, deviating at most 25 cm from the reference path.
更多查看译文
关键词
Model Predictive Control,Fleet Evaluation,Adaptive,Automatic Control,Autonomous Vehicles
AI 理解论文
溯源树
样例
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
生成溯源树,研究论文发展脉络
Chat Paper
正在生成论文摘要