Shared control leverages robotics and human strengths to enhance safety, efficiency, and task success in human-robot teams. In scenarios where communication between the human and robot is constrained by sparse, noisy, or delayed signals, effective shared autonomy requires understanding the human partner preferences and control signal characteristics, as well as robust frameworks for arbitration, control blending or autonomy allocation.
Through extensive end-user studies involving individuals with and without motor impairments, we have characterized teleoperation signals to reveal key patterns in how humans interact with robots. These findings have informed the design of shared autonomy algorithms that dynamically adapt the level of autonomy to user needs, improving performance and accessibility for diverse users. Additionally, we develop novel metrics that incorporate knowledge of high-performing human-robot teams that can be used for dynamically allocating control authority.
Through extensive end-user studies involving individuals with and without motor impairments, we have characterized teleoperation signals to reveal key patterns in how humans interact with robots. These findings have informed the design of shared autonomy algorithms that dynamically adapt the level of autonomy to user needs, improving performance and accessibility for diverse users. Additionally, we develop novel metrics that incorporate knowledge of high-performing human-robot teams that can be used for dynamically allocating control authority.
Our research demonstrates that intelligent, context-aware interpretation of human control signals—tailored to specific interfaces—leads to significant improvements in human-robot team performance. This work lays the foundation for systems that seamlessly balance human input and robotic autonomy, ensuring more intuitive and effective collaboration.