A Novel Software Architecture of Anticipatory Harvesting Robot Teams

This paper presents a novel application of anticipatory coordination and optimization principles, previously developed for autonomous inspection robot teams. The core tenet behind our research is to build a team of agile and relatively small robots, each equipped with two synergistically operating robotic arms, one arm with a soft gripper, the other arm with a set of cameras and a fruit cutting mechanism, all controlled by innovative anticipatory network (AN) algorithms. This allows the plantation management to efficiently supervise the team, coordinated and optimized by an adaptive anticipatory decision engine. Another key component of the robotic software architecture that allows the team of robots to efficiently harvest soft fruits such as strawberries or raspberries is a package of information fusion algorithms in the world model, associated with a knowledge base and a digital twin of the cover crops. We will show that AN coordination principles ensure the reliable operation of robots in case of communication disturbances or when uncertain or incomplete instructions are entered by the operator in the management and supervision system. The flexible design of the overall robotic architecture will make it possible to further develop the robot teams to respond to additional needs such as harvesting other fruits and working in various horticultural environments.