MAP2: Model- and Acceleration-Based Pursuit with MPC and Gaussian Process Residual Learning for Autonomous Racing

Key points

• Model- and Acceleration-based Pursuit (MAP) preserves the simplicity of geometric approaches while leveraging tire dynamics.1
• This paper presents MAP2, a model-based pursuit controller that combines a curvature-based kinematic MPC and a Sparse Gaussian Process (SGP) residual correction.1
• The proposed algorithm uses MPC to optimize kinematic control inputs over a prediction horizon and maps them to steering commands through a tire dynamics model augmented with SGP residual correction.1
• Compared with MAP and Pure Pursuit (PP), MAP2 reduces the average lateral tracking error by 37.99% and 44.65%, respectively, while reducing average lap time by at least 1.5%.1

Sources (1)

• [1]MAP2: Model- and Acceleration-Based Pursuit with MPC and Gaussian Process Residual Learning for Autonomous Racing
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 03:52 PM

  Autonomous racing requires accurate trajectory tracking near the handling limits of a vehicle while maintaining low computational latency.

  Model- and Acceleration-based Pursuit (MAP) preserves the simplicity of geometric approaches while leveraging tire dynamics.

Extractive summary: sentences quoted from the sources.