Flexible Collision Library

Motion planning is necessary to allow robots to move safely in real human environments while avoiding obstacles. It relies on accurate and fast collision checking to know whether poses of the robot in the world are in collision or not. Motion planners often have to deal with partial or noisy information about the environment they operate in.

During his summer internship at Willow Garage, Jia Pan worked on new motion planning approaches and a new BSD-licensed library for collision checking. Jia is currently a PhD student in Prof. Dinesh Manocha's group at UNC Chapel Hill, a world-leader in collision checking and motion planning. Jia implemented the new Flexible Collision Library (FCL), which provides the latest in collision checking capabilities including proximity information and continuous collision detection. Jia also implemented a motion planner that accounts for uncertainty and unknown space, so that planned paths are less likely to go through areas not seen by the sensors on the robot.

To find out more about Jia's work, please watch the video above and read the presentation slides. Also, read the paper here to find out more about FCL.