Human-Robot Interaction often uses Natural User Interfaces, like hand gestures, to control
robots. However, continuous direct manipulation can cause tracking noise and physical fatigue,
known as the "gorilla arm effect". To solve this, we developed a "shared autonomy" system
where the user and the robot work together. The goal is to relieve the human's physical burden
by letting the robot handle the high-precision parts of a task.
We built a simulated environment in Webots, using MediaPipe for hand tracking and an
Exponential Moving Average filter to smooth the signal. Control is managed by a Finite State
Machine: the user provides rough manual guidance, and once close to the target, the robot
autonomously executes the precision grasping and moving phases.
We conducted a within-subjects experimental evaluation with 4 participants to compare this
system against fully manual control. Quantitative results showed that the shared autonomy
system reduced Task Completion Time by 63.1% and achieved a 100% success rate.
Furthermore, subjective evaluations showed a massive drop in user fatigue and frustration. This
proves that blending human intent with robotic precision creates a highly effective and
comfortable experience.
We are excited to invite you to the ACM GreekCHI-CYCHI Symposium 2026, held on November 6-7, 2026 jointly in Athens, Thessaloniki, Nicosia and Online.