A model predictive control-based framework for time-optimal trajectory planning of fruit-picking robotic arms

Abstract

Fruit-picking robotic arms face major challenges in trajectory planning, particularly when the number of path points varies. This study proposes a trajectory planning framework for multi-degree-of-freedom robotic arms by integrating 5-7-5 polynomial interpolation with model predictive control to generate smooth and dynamically feasible trajectories. An improved Mirage Search Optimization algorithm, termed CHMSO, is introduced to enhance global optimization performance. Benchmark tests on 29 CEC2017 functions show that the proposed method improves convergence speed, population diversity, and exploration capability, with 93.97% of results outperforming comparative algorithms. For robotic arm trajectory planning, the framework generates smooth, discontinuity-free motions and reduces execution time by 54.22% compared with conventional methods. MATLAB, ROS, and physical experiments on the xArm6 robotic arm further confirm time-efficient and impact-free fruit-picking trajectories. The proposed framework is promising for precision agricultural automation and robotic applications requiring smooth, reliable, and repeatable motion execution.