AI-Based Multi-criteria Path Planning of Cartesian Robot with Telescopic Arm for Tree Fruit Picking

The paper presents a hybrid neuro-fuzzy applicative interface for path planning and multi-criteria motion optimization of an agriculture service robot with a telescopic arm designed for picking fruit from trees on the large orchards. The intelligent optimization system is based on application of artificial intelligence algorithms–neural networks and fuzzy inference system, which includes necessary knowledge about the kinematic and dynamic behavior of the robot in the operation task space, including data on the coordinates of places intended for disposal of the picked fruits from the tree. The intelligent system for path planning of the robot determines in real time the points in the working space that meet three set qualitative criteria for decision-making: (i) the shortest paths, (ii) the fastest paths, and (iii) the minimum energy consumption for catching, picking and carrying away the harvested fruits to the nearest place provided for the disposal of the fruits. The applied artificial neural network, described in the paper, is used for training, i.e. acquisition of the knowledge about the kinematic and dynamic behavior of the robot in the workspace in order to replace the process of numerically demanding calculation of the robot model in real time. Fuzzy classifier as a smart inference system has the role of choosing the best, optimal solution for the robot path that satisfies the set criteria functions.