Real-time motion intention estimation based using surface electromyography for collision avoidance

Collision avoidance has been a significant issue to guarantee human´s safety in the robot workspace. This paper presents real-time motion intention estimation based collision avoidance method using surface electromyography (sEMG). An ANN algorithm was used to estimate the upper limb motions of a subject from sEMG signals on five muscles, and the robot was controlled to keep the safety distance from the estimated motion in order to avoid the collision. The proposed method was evaluated through comparison tests with using a goniometer in real-time, and the experimental results showed a reasonable performance of collision avoidance and simultaneous response of the robot with human movements. These promising results can be useful for collision avoidance and safe human-robot interaction.