Abstract
Real-world household tasks present significant challenges for mobilemanipulation robots. An analysis of existing robotics benchmarks reveals thatsuccessful task performance hinges on three key whole-body controlcapabilities: bimanual coordination, stable and precise navigation, andextensive end-effector reachability. Achieving these capabilities requirescareful hardware design, but the resulting system complexity furthercomplicates visuomotor policy learning. To address these challenges, weintroduce the BEHAVIOR Robot Suite (BRS), a comprehensive framework forwhole-body manipulation in diverse household tasks. Built on a bimanual,wheeled robot with a 4-DoF torso, BRS integrates a cost-effective whole-bodyteleoperation interface for data collection and a novel algorithm for learningwhole-body visuomotor policies. We evaluate BRS on five challenging householdtasks that not only emphasize the three core capabilities but also introduceadditional complexities, such as long-range navigation, interaction witharticulated and deformable objects, and manipulation in confined spaces. Webelieve that BRS's integrated robotic embodiment, data collection interface,and learning framework mark a significant step toward enabling real-worldwhole-body manipulation for everyday household tasks. BRS is open-sourced athttps://behavior-robot-suite.github.io/