Robust Belief-Based Execution of Manipulation Programs
Kaijen Hsiao, Tomás Lozano‐Pérez, Leslie Pack Kaelbling · 2008
Abstract: We describe a simple approach for executing manipulation programs in the presence of significant, but bounded, uncertainty. The key idea is to maintain a belief-state (a probability distribution over world states) and to execute fixed trajectories relative to the most-likely state of the world. These world-relative trajectories, as well as the transition and observation models needed for belief update, are all constructed off-line, so the approach does not require any on-line motion planning. 1