We present new algorithms to perform fast probabilistic collision queriesbetween convex as well as non-convex objects. Our approach is applicable togeneral shapes, where one or more objects are represented using Gaussianprobability distributions. We present a fast new algorithm for a pair of convexobjects, and extend the approach to non-convex models using hierarchicalrepresentations. We highlight the performance of our algorithms with variousconvex and non-convex shapes on complex synthetic benchmarks and trajectoryplanning benchmarks for a 7-DOF Fetch robot arm.
展开▼
