TrajAudit

TrajAudit is a desktop + web app for robotics teams that need to validate motion plans before running them on real robots. It ingests trajectories from ROS 2/MoveIt, OMPL planners, or custom planners (CSV/JSON), then replays them against a digital scene to flag collisions, near-misses, joint-limit violations, excessive jerk, and timing inconsistencies. The core value is a standardized “trajectory QA report” you can attach to pull requests and releases, with pass/fail gates and regression tracking across planner versions. It’s not a magic planner replacement; it’s a practical verification layer that reduces expensive lab time and embarrassing demos. An optional AI assistant summarizes failures, suggests likely root causes (bad scene geometry, wrong frames, too-tight constraints), and proposes parameter tweaks—clearly labeled as suggestions, not guarantees.