Time-Optimal Trajectory Planning and Tracking for Autonomous Vehicles

TOTPT solves time-optimal raceline using NLP with direct collocation approach, and performs NMPC path tracking within the racing scenario implemented in Simulink.

Installation

Prerequisites

• Matlab and Simulink (tested on R2017a and R2022a on Win11)
• Install CasADi
• Install acados for Matlab

Download TOTPT and unzip at local directory

Dependencies

• Run totpt_env_variables.m to add subfolders to Matlab search path
• Open acados_env_variables_windows.m in nmpc folder, fill acados and casadi installation paths in the two lines

acados_dir = 'A:\path\to\acados';
casadi_dir = 'B:\path\to\casadi-3.6.5';

Workflow

1. navigate to tro folder, run live script tro_main to generate smoothed time-optimal raceline (1~3 min, depend on problem size)
2. navigate to nmpc folder, run nmpc_gen to generate NMPC mex function and automatically copy it to sim folder
3. navigate to sim folder, run live script racing_sim to simulate racing scenario

Folders

• racetrack-database: racelines and track widths of race tracks, forked from (https://github.com/TUMFTM/racetrack-database)
• functions: helper functions for track processing
• params: track smoothing parameters and vehicle parameters
• tro: offline time-optimal trajectory optimization framework
• nmpc: online NMPC trajectory tracking framework

Citing

@article{li2024time,
  title={Time-Optimal Trajectory Planning and Tracking for Autonomous Vehicles},
  author={Li, Jun-Ting and Chen, Chih-Keng and Ren, Hongbin},
  journal={Sensors},
  volume={24},
  number={11},
  pages={3281},
  year={2024},
  publisher={MDPI}
}