MuJoCo Robotics Lab 4: Collision-Free Robot Planning — RRT* to Execution Pipeline скачать в хорошем качестве

MuJoCo Robotics Lab 4: Collision-Free Robot Planning — RRT* to Execution Pipeline

From 35 raw RRT* waypoints down to 3 after shortcutting, then a time-optimal 2.659-second trajectory executed with 0.0037 rad tracking error. This video walks through the complete collision-free planning and execution pipeline on a UR5e in a cluttered environment. In this video: • MuJoCo as collision ground truth — using physics engine geometry for distance queries • RRT* planning in 6-DOF configuration space with obstacle avoidance • Path shortcutting — reducing 35 waypoints to 3 while staying collision-free • TOPP-RA time-optimal trajectory parameterization (2.659s execution time) • Gravity-compensated trajectory execution with 0.0037 rad tracking accuracy • Blocked-path validation — verifying the planner correctly rejects infeasible goals • End-to-end capstone demo with metrics visualization This is part of the MuJoCo Robotics Lab series — a 9-lab journey from simple planar arms to VLA-controlled humanoid manipulation, built with MuJoCo (simulation) and Pinocchio (analytical robotics). Full series playlist:    • MuJoCo Robotics Lab   Source code: https://github.com/ozkannceylan/mujoc... Notebooklm: https://notebooklm.google.com/noteboo... Tags: RRT star, motion planning, collision avoidance, TOPP-RA, UR5e, MuJoCo, Pinocchio, trajectory execution, path shortcutting, robotics implementation, obstacle avoidance