Research On Active Perception For Multi-UAVs Target Search

Unmanned aerial vehicle(UAV)that can perform related tasks in dangerous and complex unknown areas due to its air superiority,has played a significant role in industrial,medical,military and civil fields in past decades.In particular,UAV can quickly search and locate ground targets and multi-UAVs can realize rapidly target search in a large unknown environment,which has introduced new opportunities for robot coordinated control research.Focusing on how to coordinate unmanned aerial vehicle(UAV)swarm to find static and dynamic water surface and underwater targets in a complex unstructured marine environment,the active perception algorithm framework was established.Firstly,the target search model of UAVs is established based on the framework of active sensing system.Meanwhile,considering the influence of dynamic magnetic characteristics of target and environment on the search efficiency,sensor observation model is founded on the equivalent average detection width theory.Secondly,the UAVs distributed target search algorithm was built by the biologicallyinspired mechanisms called flocking and stigmergy.Furthermore,a differential evolution algorithm,which introduces adaptive operators,is proposed for the balance of exploration and exploitation in the environment to optimize the multi-UAVs collaborative search and ensure adaptability of the active sensing strategy.In this paper,maximizing the number of tracking targets in the shortest time is the global utility and the target search strategy is validated on the simulation platform to evaluate the feasibility of the algorithm in distinct unstructured scenarios with un-known spatial targets.In addition,motion control and navigation simulation of individual UAV are realized.Visualization of some dynamic environment experiments and Simulation of UAV group formation control are also completed.Simulation results show that our active perception algorithm can improve the observation perspective of UAVs in parallel and significantly reduce the redundancy of UAV group exploration.The active perception algorithm expands the number and quality of UAV group observation in time and space,which makes UAVs more efficient to complete the target search task in different types of large-scale unknown environment.