Coordination Strategies Of Multi-mobile Robot System Using Petri Nets

With the gradual developments of industrial automation technology,mobile robots have been widely adopted to transport goods in numerous industrial scenes,such as logistics industry and manufacturing industry,due to their flexibility and scalability.The Covid-19 pandemic,which exacerbates the demands for unmanned operations in industries,has facilitated the researches and applications of mobile robots.Multiple mobile robots are usually needed to work cooperatively within a limited area,forming a multi-mobile robot system.However,as the scale of the system increases,the strategy to coordinate mobile robots has become one of the most fundamental but tricky problems.This paper mainly investigates three vital issues in a multi-mobile robot system,which are path planning,collision avoidance and deadlock prevention.A multi-mobile robot system is a typical discrete-event dynamic system,which is frequently modeled by Petri nets.Therefore,in this paper,the multi-mobile robot system is modeled by colored Petri net,on the basis of which the coordination strategy is studied thoroughly.Firstly,a multi-mobile robot system which doesn’t exhibit the characteristics of place/transition interference is studied.As for avoiding collisions,a hierarchical collision avoidance algorithm based on the capacity of places and dynamic priorities is proposed,which avoids collisions by controlling both the number of tokens flowing into a place and the firing conditions of the place’s input transitions.As for preventing deadlocks,an active deadlock prevention algorithm based on reverse path subsequence,and a passive deadlock prevention algorithm based on non-cyclic places are proposed.First,deadlocks are classified into active and passive ones by its cause of formation.Next,active deadlocks caused by unreasonable order of firing transitions could be avoided by combining current places of tokens with the identification of reverse path subsequence.Then,passive deadlocks caused by path planning directly could be prevented by routing strategy and the controlling of transition firing sequence.Finally,an unlocking strategy based on interim places,as well as an interim place filtering algorithm is proposed.Secondly,a multi-mobile robot system that exhibits the characteristics of place/transition interference is studied.First,the judgment of place/transition interference is achieved by dividing the map into a set of grids.Next,in order to reduce the computational complexity,the concept of augmented contour grids is proposed,and their fast searching algorithm is designed based on the sampling of Bézier curve and mobile robots’ contours.Then,a collision avoidance algorithm based on augmented contour grids,which avoids collisions when facing place/transition interference,is proposed.Finally,an active deadlock prevention algorithm based on reciprocal interfering path subsequence is designed,and the passive deadlock prevention algorithm is improved on the basis of augmented contour grids.Lastly,we focus on the mixed traffic system which contains both mobile robots and human-driven vehicles,collectively called “moving objects”.The coordination strategy of a mixed traffic system,where there exist mobile robots that do not possess intelligence,is studied.First,the moving objects of a mixed traffic system have been classified into fully-controllable ones,partially-controllable ones,and fullyuncontrollable ones according to their respective controllable degrees.Then,based on the behaviors of moving objects with different controllable degrees,the concepts of partially-controllable risk area and fully-uncontrollable risk area are proposed.Finally,a collision avoidance strategy based on risk area,as well as a deadlock prevention strategy based on controllable degree and reverse path subsequence,is proposed.The three coordination strategies mentioned above guarantee that the system is absolutely collision-free,while boosting the operational efficiency by preventing deadlocks.Simulated experiments and industrial applications both demonstrate the effectiveness and efficiency of the strategies.Meanwhile,the proposed strategies can meet the real-time control requirements of an industrial system,which contributes to both academic research and engineering applications.