Proprioceptive-visual correspondence enables self-other distinction in humanoid robots

A humanoid robot was developed that learns self-other distinction through proprioceptive-visual correspondence, eliminating the need for identity labels or kinematic models. This approach enables the robot to create a predictive self-model that maps joint configurations to 3D body occupancy, facilitating tasks such as target reaching and collision-aware motion planning in multi-agent environments. This advancement is significant for practitioners as it enhances robots' social intelligence and coordination capabilities in shared workspaces.