Learning-Deep-Learning

Baidu Apollo EM Motion Planner

June 2024

tl;dr: An optimization-based motion planner with path-speed decoupled method.

Overall impression

Apollo EM motion planner is a scalable and easy-to-tune framework to handle traffic rules, obstacle decision and smoothness.

EM planner significantly reduces computational complexity by transforming a three-dimensional station-lateral-speed problem into two two- dimensional station-lateral/station-speed problems.

The paper does have some drawbacks, such as limited discussion on handling highly dynamic environments with unpredictable behaviors.

Key ideas

Technical details

Notes