Real-time informed path sampling for motion planning search

Mobile robot motions often originate from an uninformed path-sampling process such as random or low-dispersion sampling. We demonstrate an alternative approach to path sampling that closes the loop on the expensive collision-testing process. Although all necessary information for collision testing a path is known to the planner, that information is typically stored in a relatively unavailable form in a costmap or obstacle map. By summarizing the most salient data in a more accessible form, our process delivers a denser sampling of the free path space per unit time than open-loop sampling techniques. We obtain this result by probabilistically modeling-in real time and with minimal information-the locations of obstacles and free space, based on collision-test results. We present CALM, the combined adaptive locality model, along with an algorithm to bias path sampling based on the model's predictions. We provide experimental results in simulation for motion planning on mobile robots, demonstrating up to a 330% increase in paths surviving collision test.