Decision-Making and Planning – Benchmark Models

This page brings together Benchmark Models contributed by the community, following a similar format. A Decision-Making and Planning project generally includes README.md (full deployment and result reproduction guidance), dataset-raw data (raw data and processing code), dataset-processed data (standardized processed data), env (autonomous-driving simulation environment), model (decision-making and planning models), evaluator (result evaluation), and utils (utility functions and supporting code).

VCDI

  • Description: Model autonomous-vehicle trajectory planning using the Waymo dataset
  • Overview: This model uses multimodal Transformer features to combine action dependencies between the autonomous vehicle and surrounding road users with map-scene information. A Gaussian-distribution-based cost-function module then optimizes the trajectory planning result.
  • Link：https://github.com/YNYSNL/VCDI

TRPPO

  • Description: Use a PPO algorithm with a Transformer architecture for behavior decision-making
  • Overview: Transformer-enhanced PPO improves the ability to capture long-term dependencies in complex traffic scenarios. Better awareness of global contextual information substantially improves safety and robustness in multi-vehicle scenes, supporting complex operations such as lane changes and emergency avoidance.
  • Link：https://github.com/TOPSlearningcenter/TRPPO