sharedcontrol

Shared control is a paradigm in human-robot collaboration in which control authority is distributed between a human operator and an autonomous controller. Rather than relinquishing control entirely to either the human or the machine, the system continuously blends or arbitrates actions so that both contribute to the execution of tasks. This approach is designed to leverage human judgment and situational awareness while benefiting from machine precision, speed, and reliability.

Common elements of shared control include intent inference, context awareness, and adaptive arbitration. Intent inference attempts to infer the user’s goals from inputs or observations, while adaptive arbitration determines how much influence the human or the autonomous system should have at any moment. Control can be blended, by weighting and summing inputs from both agents, or switchable, by dynamically changing control modes. Interfaces for input range from traditional devices like joysticks and pedals to gaze tracking, voice, or haptic feedback, and state information from the environment informs both agents.

Benefits of shared control include reduced cognitive and physical workload, increased safety and robustness in uncertain environments, and maintained user involvement and trust. Challenges involve accurately interpreting user intent, managing latency and synchronization, calibrating the degree of autonomy, ensuring transparency, and preventing over-reliance on automation.

Applications span assistive devices (such as powered wheelchairs and exoskeletons), surgical robotics, driving and aerospace systems, teleoperation, drones, and industrial robotics. Evaluation typically considers task success, completion time, error rates, operator workload, and trust or comfort with the system, guiding improvements in arbitration strategies and interface design.