Research on Low-Cost Robotic Arm Intelligent Control System Based on Heterogeneous Computing and Depth Perception

This study proposes a low-cost intelligent control system for robotic arms that integrates depth perception and heterogeneous computing, specifically for automotive parts sorting applications. A three-dimensional visual servo model is developed using AstraPro, and dynamic path planning is optimized with the A* algorithm. Additionally, adaptive adjustments for electromagnetic suction are implemented through PID-PWM control, leveraging a dual-core synergistic control architecture that combines Raspberry Pi and STM32. Experimental results demonstrate that the system achieves a positioning accuracy of ±0.8 mm, reduces the path planning response time to 45 ms, and attains a success rate of 92% or higher for gripping shaped parts. Remarkably, the overall cost of the system is only $800. This research offers a cost-effective intelligent sorting solution that enhances automation in the automotive industry.