A Simulation-based Approach for Evaluating Shared Control Algorithms for Mobile Robots
Ahmet Saglam, Yiannis E. Papelis · 2024
Shared control algorithms in teleoperated robots blend user commands with robot intelligence. However, performance evaluation of these algorithms is challenging due to reliance on human feedback and limitations of physical test environments. To overcome these challenges, we propose a simulation approach for evaluating the effectiveness and efficiency of a shared control algorithm designed for mobile robots in realistic scenarios. A set of performance metrics is introduced to quantitatively assess the performance of the algorithm. Monte Carlo techniques are used to assess the robustness of the algorithm by running batch simulations in a fixed virtual world where both obstacle configurations and user inputs vary randomly across simulation scenarios. Preliminary results from various experimental setups using our blended control algorithm demonstrate its potential effectiveness. We aim to understand how well a shared control algorithm performs in environments that closely mirror real-world conditions while avoiding logistical constraints associated with human subject experimentation.