Propagation of temporal and pose uncertainty for true mobile manipulation

Conference: ISR/Robotik 2014 - 45th International Symposium on Robotics; 8th German Conference on Robotics
06/02/2014 - 06/03/2014 at München, Germany

Proceedings: ISR/Robotik 2014

Pages: 6Language: englishTyp: PDF

Personal VDE Members are entitled to a 10% discount on this title

Authors:
Ruehr, Thomas (KUKA Laboratories GmbH, Germany)
Beetz, Michael (Institute for Artificial Intelligence, University of Bremen, Germany)

Abstract:
Pose uncertainty is ubiquitous in mobile manipulation. It’s important to realize that both positional and temporal errors will ultimately result in positional misalignment when objects are in motion. In order to judge how given translational, rotational and temporal variance affects a task at hand, they are combined and propagated to a coordinate frame that is relevant to the task. Typically, approaches for uncertainty propagation treat the world as quasi-static. We are presenting a framework that treats time adequately by two additions. Firstly, queries that span over a time interval are supported, such as ’where would object Oa that was observed at time T0 be located with respect to the robot pose at time T1 and what is its pose belief distribution’. Secondly, uncertainty in time is supported, such as when we need to reason about the probability distribution of an object that was observed with a sensor that doesn’t provide accurate timestamps. The library is open source, fully integrated with the ROS ecosystem and offers simple to use and performant interfaces.