Research on design and experiment of rear-drive apple harvesting Robotic arm based on obstacle avoidance posture conditions

• Designed a rear-driven apple-picking robotic arm • An obstacle-avoidance posture evaluation function is proposed to determine the optimal picking posture angles for collision avoidance. • Validation experiments on the obtained posture angles demonstrated success rates of 82.21% for the upward picking angle of 43° and 57.9% for the downward picking angle of 49°. Aiming at the problems of small space coverage of fruit area and high equipment price in apple picking operation using traditional industrial manipulator, this paper designs a driving rear-mounted apple picking manipulator based on obstacle avoidance posture condition in combination with the growth environment and picking demand of fruit. Based on the investigation of the environment of Red Fuji orchard in Weihai, Shandong Province, a design scheme of driving rear-mounted apple picking manipulator suitable for large space is proposed, and the modular design of each part of the manipulator is carried out. Aiming at the problem that the horizontal picking attitude of the three-claw end-effector is easy to interfere with the 'connecting branch', an obstacle avoidance attitude evaluation function is proposed to solve a set of adaptive obstacle avoidance picking attitude angles, which are 43° for upward picking and 49° for downward picking. Finally, the designed mechanical structure is processed and assembled, and the motion control system of the manipulator is designed based on the CANopen communication sub-protocol DS301 and the control system sub-protocol DS402. The final picking test demonstrated a success rate of 82.21% for upward picking with an obstacle avoidance posture angle of 43° and 57.9% for downward picking with an obstacle-avoidance posture angle of 49°, validating the rationality of the obstacle avoidance posture angles. In multi-fruit picking tests, the overall success rate reached 86%, with an average harvesting rate of 11.5 seconds per fruit, achieving continuous harvesting functionality for the robotic arm.