An Improved Multi-Objective Trajectory Planning Algorithm for Kiwifruit Harvesting Manipulator

Trajectory planning is always a hot issue for harvesting manipulators, considering the limitations of the manipulator mechanical structure and other nonlinear factors that result in long time, big jerk and high energy consumption of the manipulator in tracking the picking path. The cubic spline algorithm, fifth-order polynomial interpolation algorithm, a fusing cubic spline and fifth-order polynomial algorithm aim to shorten harvesting time, reduce jerk and enhance the robustness of the harvesting manipulator, respectively. Then the trajectory planning problem is converted into a multi-objective optimization solution problem. A non-dominated sorting genetic algorithm-the third version (NSGA-III) technology is used to address multi-objective optimization constrained problems with time, energy consumption and jerk optimization objectives. The Pareto optimal set containing the multiple constraints is obtained through the NSGA-III algorithm. Compared to a non-dominated sorting genetic algorithm-the second version (NSGA-II) technique, optimal solution with a shorter time is chosen from the non-dominated Pareto optimal set and sent to the manipulator controller actuator. Under the robot operating system (ROS), the cubic spline algorithm, fifth-order polynomial interpolation algorithm, a fusing cubic spline and fifth-order polynomial algorithm with the structural coefficient obtained respectively through NSGA-III algorithm and NSGA-II algorithm are simulated. The simulation results show that the NSGA-III algorithm has a better result in addressing the trajectory planning problem. The experimental analysis of the harvesting manipulator verifies the proposed scheme that the cubic spline algorithm based on the NSGA-III algorithm enables the harvesting manipulator to track the picking path faster and smoother, which is effective and feasible for reducing the time and improving the harvesting efficiency of kiwifruit.