Stability tests of two-finger tomato grasping for harvesting robots

In this study, the theories of spatial and contact grasp stability were extended and integrated into a whole system, and then a vision processing approach that extracts the relevant information for synthesising plate and curved finger grasps for unknown tomato fruits from tomato images was presented. Finally, stability tests involving grasping tomatoes with two parallel fingers were performed using two types of fingers (plate and curved fingers). Existing theories of grasp stability related to rigid objects could be integrated and extended to analyse the grasping stability for half-ripe tomatoes. Curved fingers were more suitable for stably grasping tomatoes than were plate fingers. The prediction method of stable grasp regions can be regarded as a potential strategy (algorithm) for achieving a programmed control of two-fingered tomato grasp stability based on vision feedback. Visual perception is used to reduce the uncertainty and obtain relevant geometric information about the tomatoes during harvesting.