Mobile Manipulation

Mobile manipulation coordinates navigation, perception, and physical interaction so robots can reach objects, choose workable base positions, and perform tasks while moving between locations. Approaches range from geometric mapping, reachability analysis, motion planning, and whole-body model-predictive control to imitation learning, reinforcement learning, and vision-language-action policies trained on robot or human demonstrations and simulation data. Work increasingly combines these methods with hierarchical skills, spatial memory, active perception, and recovery for long-horizon tasks in homes and other changing environments. A central challenge is transferring success from individual skills and simulation to reliable real-world behavior under shifts in viewpoint, base pose, clutter, contact, and disturbances; aggregate benchmark scores can obscure these weaknesses.