Robot arm collision detection using joint position and torque controller

Utilizing the MuJoCo simulation platform, I developed experimental environments for robotic arms, enabling advanced testing and analysis. The core of my research involved proposing an transformer-based model for collision detection in robotic arms. This model’s unique feature is its ability to function without relying on external sensors, instead using only joint position and torque controller data. This approach potentially simplifies robotic systems while maintaining high accuracy in collision detection.