Application of an Attention-Enhanced PID-LSTM Hybrid Control Algorithm for Industrial Robotic Arm Trajectory Tracking
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
To address the challenges of insufficient trajectory tracking accuracy, delayed dynamic response, and weak anti-interference capabilities faced by industrial robotic arms in precision manufacturing scenarios, this paper proposes an attention-enhanced PID-LSTM hybrid control algorithm. This algorithm employs a three-layer collaborative architecture consisting of "PID basic control + LSTM error prediction + attention optimization and adjustment." The PID module uses an incremental structure (parameters Kp=2.8, Ki=0.35, and K_d=0.72) to ensure initial stability. The LSTM module (5 nodes in the input layer and 16 nodes in the hidden layer) predicts compensation based on historical error sequences. The attention module filters key information using error gradients and a temporal decay factor (α=0.95). Using a UR10 industrial robot arm (with a repeatability of ±0.02mm) as the control object, a simulation platform with a multibody dynamics model was built in MATLAB/Simulink. Experiments were conducted using three trajectory configurations: a straight line (0.3m), a circular arc (radius 0.15m), and a B-spline curve (cubic interpolation). ±8% load fluctuation and 0.2V Gaussian noise were introduced to simulate disturbances. Simulation results show that for the B-spline trajectory, the hybrid algorithm achieved a mean position error of 0.023mm, a 72% reduction compared to the traditional PID (0.082mm) and a 52.1% reduction compared to the simple LSTM (0.048mm). The step response time was 0.15s (44.4% shorter than the traditional PID), and the overshoot was 1.8% (78.8% lower than the traditional PID). The error change rate under an 8% load disturbance was 6.7%, only 31.5% of that of the traditional PID. This algorithm balances multi-dimensional performance and is adaptable to precision scenarios such as electronics manufacturing and automotive parts assembly.
Fields of Study & Subjects
Advanced Algorithms and Applications
Adaptive Control of Nonlinear Systems
Robotic Mechanisms and Dynamics
Publication details
- Journal / source
- IEEE International Conference on Power, Electronics and Green Energy (ICPEGE)
- Pages
- 208–213
- Publication date
- 28 January 2026
- Language
- EN