A quicker eye for robotics to help in our cluttered, human environments
Publication/Creation DateMay 23 2019
DescriptionIn preparation for introducing robots to more clustered and chaotic environments, researchers at the University of Michigan developed the Pull Message Passing for Nonparametric Belief Propagation algorithm to hasten machine perception time. Using a Fetch robot, the team demonstrated that the algorithm could reduce the time to accurately determine an object's pose down to 10 minutes, an action that previously took over an hour and a half. This new algorithm uses data from sensors to "pull" the robot to the correct answer, whereas previous algorithms "pushed" data to the robot over several iterations.
Date archivedJune 17 2019
Last editedJune 24 2019