Multi-sensor fusion control technology and its automation implementation methods for electromechanical servo systems with nonlinear friction

Engineering notes

Key topics: autonomous driving, sensor fusion, multi-sensor fusion, deployment, perception, control. See the paper for implementation details and experimental results.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为端到端自动驾驶、BEV感知、3D目标检测、轨迹预测、路径规划、LiDAR感知等高价值论文补充中文说明。

Original abstract

Introduction Complex mechanical systems are affected by multiphysics coupling, nonlinear friction, and time-varying disturbances, making high-precision automated operation difficult. Directly mapping heterogeneous sensing data to real-time control actions remains challenging. Methods This study developed a multi-sensor fusion and adaptive control framework (AFCF) by integrating a Dual-Stream Attention Mechanism (DS-AM) with an improved Twin Delayed Deep Deterministic Policy Gradient (TD3) controller. DS-AM decouples high-frequency vibration and low-frequency current features, while prioritized experience replay and dynamic constraints improve learning efficiency and torque-execution safety. Results At 5 dB noise, AFCF achieved a tracking RMSE of 0.035 mm. For a variable-curvature butterfly trajectory, it limited the maximum contour error to 4.8 μm and estimated surface roughness to 0.52 μm. Under intermittent impact, it reduced energy consumption by 9.62% and peak mechanical acceleration by 45.6%. Discussion AFCF integrates heterogeneous perception and adaptive control in a closed loop, supporting robust and energy-aware servo control under complex conditions. Further validation with physical hardware and lower-complexity models is required for broader deployment.

Links and sources

• PDF from original source

Comments