Adaptive Multimodal Control of Robotic Arm

Adaptive algorithm using vision, force & IMU with DRL achieves ±0.1 mm accuracy, 42.1% better stability & 31.5% faster tasks.

The paper focuses on enhancing precision control of robotic arms operating in complex and dynamic industrial environments. It introduces an optimized adaptive algorithm that combines multimodal perception (vision, force, IMU) with deep reinforcement learning (DRL) and edge computing. The proposed system achieves ±0.1mm positioning accuracy, improves disturbance suppression by 42.1%, and reduces task execution time by 31.5%, outperforming traditional PID and fuzzy control systems.