Collision Avoidence And Optimal Coordination In Multiple Robots Symstems

Compared with the system comprised of single robot, cooperation and coordination between each single robot in multiple robots system increase the capacity, working efficiency, flexibility and robustness of the whole system. Therefore, multiple robots systems have emerged and become one of the hotspots and difficult issues in robotics. On one hand, cooperation and coordination between robots nonlinearly and greatly improves performance of the multiple robots system, it also increases complexity of the system significantly. Issues involved in the multiple robots system research include path/trajectory planning, collision; detection, collision avoidance strategies, coordination strategies and so on. But a problem running through them is conflict avoidance and conflict resolution. One of the most fundamental and important conflict is the collision conflict between each single robot, and collision conflict between robots with the given environment. Collision avoidance in multiple robots system provides the premise and lays the foundation to ensure safety and efficiency.To avoid the possible collisions in multiple robots systems with shared workspace, one method is to seek collision-free paths for each robot through coordinated path searching methods such as artificial potential field or probabilistic maps. Through centralized or decoupled coordination control strategy is another way. Due to large numbers of DOF, complex movement pattern and numerous components of industrial robots, current research about coordinated path searching for collision avoidance strategies and coordinated control methods in multiple industrial robots systems are not complete, generally limited to dual-arms robot or dual-robot system. Therefore studying collision avoidance strategies and coordinated control algorithms for multiple industrial robots have practical significance and value of engineering.In terms of collision avoidance and optimal coordination in multiple robots system, the collision detection algorithm, collision avoidance and coordination strategies were proposed as two main sections respectively in this paper:(1) An overview on collision detection algorithms based on separation distance was summarized. Firstly analyzed the process flow of the multi-body collision detection algorithms, which is widely applicable in robot, NC machine and other fields. According to the result whether returnning the minimum distance between objects or not, the collision detection algorithm was classified into collision detection algorithms based on separate distance (proximity query) and collision detection algorithms which is not based on separate distance. Then a survey of collision detection algorithms based on separate distance according to bounding volume hierarchies, algebraic geometry, geometrical features, simplex, mathematical programming and distance fields on the basis of the basic principles these algorithms adopted computing the separation distance in narrow phase was classified and discussed. Further, conclusions were conducted that the objects for collision detection algorithms are developing towards multi-representations, non-convex, multi-objects and collision detection algorithms are developing towards dynamic detection, continuous detection, and parallel computation. Finally, some urgent problems for collision detection algorithms were pointed out.(2) A second order cone programming based convex superquadrics separation distance computation algorithm was proposed. It treated descriptive primitives convex superquadrics of complex object as the study objects. The separation distance between two convex superquadrics was written as a general nonlinear programming (NLP) problem with constraints with fractional exponents, then this NLP was transformed into an SOCP problem with the algorithm that transforms constraints with fractional exponents into intersection between a set of second order cone and a set of linear constraints. Finally, a new primal-dual predictor-corrector interior point method was used for solving the SOCP problem, hence achieved exact proximity queries between convex superquadrics. The proposed algorithm is particularly suitable for proximity queries in narrow phase of static collision detection algorithms, and achieved continuous collision detection between rectilinear translation superquadrics.(3) A temporal displacement strategy with a delay start was proposed to achieve collision avoidance and optimal coordination for multip robots systems. It treated multiple industrial robots system (not limited to two) in the working environment with given trajectories as the sdudy case. With the definition of standardized path and standardized trajectory, the spatial collision information between each link was reflected as the time collision information. The sufficient conditions for collision free and the necessary conditions for optimal coordination in multiple robots system were deduced adopting a temporal displacement strategy with delay start. Then two mixed integer linear planning models based on the conditions were established respectively. Finally correctness of the proposed algorithm was validated through six sets of simulation experiments in dual-robot system and a set of simulation experiment in triple-robot system.(4) A simulation software for multiple robots system was estabished. Firstly, the architecture and the process flow of the simulation software for multiple robots system were planned. Then the OpenGL framework in MFC for visualization was configurated, meanwhile it was the basis for embedding collision detection algorithm, motion planning algorithm and collision avoidance coordination algorithm. Finally, a virtual teaching interface was developed, providing a relatively good user-friendly graphical user interface.