Research on Trajectory Planning of Loaders Based on Gaussian Process Regression and Particle Swarm Optimization

Bucket trajectory planning is a core technology for the automated and intelligent operation of loaders. To solve the problem of trajectory planning during the bucket operation and the bulk material transportation process, this study proposes a bucket trajectory planning method based on spline curves. First, a kinematic model of the loader’s working device is established to analyze the relationship between the bucket tip trajectory, bucket rotation angle, vehicle displacement, and the stroke of the lift cylinders and tilt cylinders. On this basis, trajectory planning schemes for different bucket depths are formulated, and the effectiveness of the approach is verified through Adams-EDEM (Automatic Dynamic Analysis of Mechanical Systems-Engineering Discrete Element Method) co-simulation using the measured parameters of the bulk material pile. Subsequently, a Gaussian Process Regression (GPR) model is constructed to evaluate the loader’s bucket performance. Finally, the parameters of the bucket trajectory are optimized using Particle Swarm Optimization (PSO) with the application of a multi-objective optimization strategy and a two-stage nested optimization strategy. The results show that the two-stage nested optimization achieves a more significant improvement: the filling rate increases by 7%, and the bucket resistance decreases by 24.8%. Compared with multi-objective optimization, the two-stage nested optimization increases the filling rate by 2% and reduces the bucket resistance by 6%. Compared with Genetic Algorithm optimization, the full bucket rate is increased by 5%, and the digging resistance is decreased by 8%.