HITL-D: Human In The Loop Diffusion Assisted Shared Control
Researchers have developed HITL-D, a new shared control framework that combines human input with diffusion-based AI policies for robotic manipulation. This system aims to improve user performance in complex tasks by providing autonomous updates to end-effector orientation, reducing the need for extensive joystick control and lowering mental workload. A user study with 12 participants showed HITL-D reduced task completion times by 40% and perceived workload by 37% compared to traditional teleoperation. AI
IMPACT This framework could enhance human-robot collaboration in complex manipulation tasks, potentially improving efficiency and reducing operator fatigue.