MPTREE: Motion primitive tree exploration for trajectory planning with dynamic obstacle avoidance
Industrial Engineering Day, University of Trento, 2022
The MPTREE is a motion primitive tree exploration algorithm for trajectory planning with dynamic obstacle avoidance. The algorithm is based on the exploration of the vehicle state space by means of a tree. The algorithm is able to generate trajectories in real-time and to avoid dynamic obstacles.
The motion planning and decision-making algorithms have to satisfy the following high level goals:
- Generate trajectories (path plus speed profiles and actuation commands) to navigate autonomously in complex urban scenario.
- Consider communicated by other vehicles or use advance motion prediction models of other road users.
- Change the type of maneuver generated according to different vehicle categories (e.g. cars vs trucks) and according to different safety and comfort criteria.