Optimal Image-Based Task-Sequence/Path Planning and Robust Hybrid Vision/Force Control of Industrial Robots

In this paper, a novel image-based task-sequence/path planning scheme coupled with a robust vision/force control method is suggested for solving the multi-task operation problem of an eye-in-hand (EIH) industrial robot interacting with a workpiece. The proposed method suggests an optimal task sequence planning scheme to perform all the tasks and an optimal path planning method to generate a collision-free path between the tasks when the robot performs free-motion. To this end, a new method is presented which solves both problems simultaneously. A novel deadlock-free modified artificial potential field (MAPF) based on rotational potential force is developed for generating the collision-free path betweentasks in the three-dimensional (3D) environment. The parameters of the MAPF and the sequence of the tasks are found by an optimizer simultaneously. This problem can be considered as a MAPF-constrained-generalized-traveling-salesman-problem (MAPF-CGTSP), which is a mix-integer optimization problem. The mix-integer version of multi-tracker optimization algorithm (MTOA) is developed to solve the problem. However, since image-based visual servoing (IBVS) is used for motion control, the planning is conducted in the image space. Integrated with the proposed planning method, a novel chattering-free filtered quasi sliding mode controller (FQSMC) is specially designed for robust vision/force control of the robot. FQSMC exploits a novel variable-gain orthogonal-sliding-manifold (VGOSM)which enables the robot to switch between free-motion mode and interaction mode. FQSMC overcomes large uncertainties and filters out the existing noises by exploiting an intrinsic filter within its control law. Experimental results show the superiority of the proposed approach to other state-of-the-art methods.