Research On Distributed Cooperative Area Searching Of Multiple Unmanned Aerial Vehicles

Cooperative surveillance of multiple Unmanned Aerial Vehicles (UAV) will be the main manner to get battlefield information in the future, and cooperative area searching is also a very important operation style for multiple UAVs. In recent years, how to control a team of UAVs to perform target searching mission in the complex environment effectively has received significant attention of cooperative control community, it is of great theoretical and practical value both in military and civilian areas. This dissertation studies the environment model, distributed decision-making method and online path planning and coordination method for multi-UAVs cooperative searching problem. The main work and contributions are as follows:(1) The elements that influence the multi-UAVs cooperative area searching problem were analyzed, and a distributed framework to this problem was presented.Following thorough analysis on the basic search problem and multiple UAV search problem, the factors including UAV platform, airborne image sensor, targets and network communication are summarized, and formalized models were established respectively. Then, based on the OODA (Observe, Orient, Decide and Act) decision process model, we established a hierarchical process for single UAV autonomous search. Furthermore, a distributed architecture for multi-UAVs cooperative search was proposed. Based on these, the multi-UAVs cooperative searching problem was divided into three key sub-problems, which are cooperative search environment modeling, distributed search decision-making and online path planning and coordination. (2) The extended search map based search environment description model wasproposed, and compensation method was presented to the information non-coherence problem under global communication delay during the search map updating.Based on basic search map, the environment description model was discussed in detail. A new search map initialization method based on target distribution was introduced. Then we employed Bayesian principles to analysis the problems of observation updating, communication updating and prediction updating of the search map, and proposed an extended search map (ESM) model based on digital hormone diffusion mechanism. Simulation results show that the ESM combines the validity of certainty map with the coordination ability of hormone, and can provide an effective means for cooperative search problem. Furthermore, the problem of information coherence for multi-UAVs cooperative searching was presented for the first time, the key elements and reasons which cause the information non-coherence were discussed and analyzed. Then to reduce the information error under global communication delay, two information non-coherence compensation methods were put forward, which are delay time based Dead Reckoning method and Kalman estimation based method. Simulation results demonstrate that these two methods can estimate the communication delay state effectively, and make a good compensation to the information errors of different UAVs'search map.(3) The distributed receding horizon (RH) optimization model for multi-UAVs cooperative search decision problem was established. A Nash optimality and PSO based iterative algorithm was proposed to the decentralized RH optimization problem solving.The multi-UAVs is a dynamically decoupled system where the task couple the dynamical behavior of the systems. Focusing on this kind of decoupled system, we discussed a distributed system modeling method based on the idea of distributed model predictive control (DMPC). Then the multi-UAVs cooperative search decision problem was posed as an optimization problem, and a distributed RH optimization model was established. In our model, the centralized online optimization problem for the whole system was decomposed into a set of decentralized optimization problems for single UAV subsystems, which greatly reduce the size of online optimization problem. Then, to the decentralized optimization solving, a Nash optimality and Particle Swarm Optimization (PSO) based method was presented. Simulation results demonstrate that our distributed method is a sub-optimal one as compared with centralized optimization method; it can bring down the decision-making time effectively and reduce the complexity of online computing. Furthermore, by the comparison simulation with several other classical search methods, such as Greedy search, Random search and Scan-line Search, our methods can make multiple UAVs to explore the area with greater target probability in a shorter time, and help them find more targets.(4) A modified Rapidly-exploring Random Tree (RRT) based online path planning algorithm was presented. A state prediction based path coordination algorithm for multiple UAVs collision avoidance was proposed.To satisfy the real-time and safety requirement of multi-UAVs cooperative search path planning, a hierarchical autonomous path planning framework was put forward. Under this framework, the original problem was divided into two layers, one was the single UAV online path planning problems, and the other was multiple UAVs collision avoidance path coordination problem. Then, by introducing the heuristic information and improving the extension of node to basic RRT, a modified RRT based UAV path planning algorithm was proposed. Simulation results demonstrate that the modified RRT method is real-time and effective planning algorithm, it can obtain feasible solutions quickly for UAV path planning when the environment or tasks are dynamically changed. Finally, to the multiple UAVs collision avoidance path coordination problem in the same air area, we established a distributed RH optimization model and presented a state prediction based path coordination algorithm. Several simulation results show that our algorithm can bring a within-network collision avoidance capability for multiple UAVs, it can achieve path coordination effectively on condition that there are communication link between two conflict UAVs.