HomeLearnCoursesHackathonsAccount
Motion Planning & Trajectory Optimization
Real-Time Replanning and the Speed-Quality Tradeoff · 1/2

Static assumptions break down in real environments

shelving above, robots and planned routes below

A motion plan computed once in advance assumes the environment is static and fully known at planning time, every obstacle accounted for, nothing moving. Real environments rarely cooperate with that assumption. A person walks into a warehouse aisle, another robot changes course, a door that was open is now closed. When the world changes after a plan was made, the plan can become invalid or unsafe, and the robot needs to detect that and replan, ideally before it collides with anything.

This requirement, replanning quickly in response to a changing environment, is a different problem from planning well in a static environment, and it pulls in a different direction. A thorough planner might spend real computation time searching for a better path or optimizing an existing one, but that thoroughness becomes a liability the moment the environment is changing faster than the planner can keep up.