Simulation model of an intelligent control system for the agricultural grip manipulator based on the multi-agent neurocognitive architectures training

The problem of increasing the efficiency of production and harvesting of fruit, vegetable and berry products has led to the need to develop technologies for unmanned harvesting of fruit and vegetable products. The most important task in the design of an agricultural robot is the task of automatic non - damaging separation of products from plantings of fruit crops. The work presents a prototype of an anthropomimetic manipulator, which is designed in the form of a human hand. A simulation model of an intelligent control system for the grip of an agricultural anthropomimic manipulator has been developed. The control system is a distributed adaptive learning automatic system (intelligent agent) based on a multi-agent neurocognitive architecture. The process of acquiring by an intelligent agent the knowledge necessary for studying the system "robot manipulator - object of influence" is presented, in particular, the process of setting and tracking the angle of rotation of the gripper fingers articulation, as well as controlling the tension of cables and the load of motors according to data from the sensor system of an agricultural robot.
The results of this study can be used to develop control systems for the manipulator of autonomous robots and robotic systems for agricultural purposes.