Research On Multi-Target Search Method Of Swarm Robots In Unknown Obstacle Environment

Target search is widely used in post-disaster search and rescue,wildfire monitoring,enemy reconnaissance,and other situations,and the swarm robot system is more efficient than a single robot in target search.The swarm robot system is an artificial system that is inspired by the group behavior of social creatures in nature and consists of multiple robots with simple information processing functions.Making full use of the high efficiency,flexibility,scalability,and robustness of the swarm robot system in target search can greatly reduce the time for searching targets,thereby reducing the loss caused by excessive search time.This paper focuses on the multi-target search of swarm robots and studies the multi-target search method of swarm robots in the environment of unknown obstacles.The main research work and research results are as follows:(1)It is established that a multi-target search of swarm robots in an unknown obstacle environment model,and a multi-target search algorithm framework in an obstacle environment is constructed based on the finite state machine of the robot.In the framework,the algorithm is divided into four parts: obstacle avoidance control strategy,collaborative search algorithm,roaming search algorithm,and task allocation strategy.(2)It is considered that the robots can distinguish the target according to the target signal,in the target signal model.For the obstacle avoidance control strategy,if obstacles avoidance is inevitable,the avoidance situations are divided into continuous obstacles and discontinuous obstacles,and the avoidance direction is divided into counterclockwise obstacle avoidance and clockwise obstacle avoidance.When the robots calculate the speed,the obstacle avoidance situation and the historical obstacle avoidance direction are comprehensively considered,and then the boundary scan obstacle avoidance strategy is proposed.For the collaborative search algorithm,the target position estimation is combined with the particle swarm optimization,and a target position estimation particle swarm optimization is proposed.Combining the above method with simplified virtual-force model and response threshold-based task allocation strategy,a multi-target search algorithm under the signal model with target distinguishable is constructed.And simulations verify the efficiency of the algorithm in static multi-target search.(3)The target signal model is improved.In the target signal model,the robot cannot distinguish the target according to the target signal,and the target is moving.For the task division strategy,the distance between individual robots and the target signal are regarded as two influencing factors of task allocation strategy,and a distance-based dynamic task allocation strategy is proposed.For the roaming search algorithm,the search area is gridded,and the confidence area pheromone,which represents the confidence area in numerical value,is proposed.For the collaborative search algorithm,a probability update form is introduced into the update of the previous best position and the global best particle of the particle swarm optimization,and the probabilistic finite particle swarm optimization is proposed.For the obstacle avoidance control strategy,the boundary scan obstacle avoidance strategy is improved into a strategy adapted to the grid environment.On these basis,a multiple motion target search algorithm is constructed in the target indistinguishable signal model.The simulation results show that the algorithm has superior performance for searching multiple motion targets.In summary,this paper studies the multi-target search method of swarm robots in the environment of unknown obstacles.The multi-target search algorithm is constructed as an algorithm framework consisting of four parts.Based on the algorithm framework,the multiple static target search and multiple motion target search are studied,and the corresponding algorithms are proposed respectively.And the superior performance of the algorithm in the corresponding environment is verified by simulation.