Research On Physicomimetics Approach For Formation Control And Reconfiguration Of Unmanned Aerial Vehicles

There are many advantages of UAV formation over a single UAV,such as,more flexible,more fault-tolerant,and more cooperative.Hence,UAV formation is an important direction of the application of UAVs in the future.Driven by the requirements,the problem of formation control and reconfiguration has become a hot topic.However,it is difficult for traditional approaches to simultaneously satisfy the requirement of scalability,flexibility and real-time computation.More works are required to narrow the gap between theory and application.Motivated by the above considerations,we investigated practical and flexible approaches for UAV formation control and reconfiguration in the dissertation.We focused on the problem from a novel view of physicomimetics.The main work and contribution include:The problem was formulated and decomposed from the perspectives of time and control hierarchy.The reconfiguration problem was divided into 3 different categories:the reconfiguration problem due to mission update,UAV adding or lost and environmental change.The problem was divided into different subproblems,including,low-level control of UAV,trajectory tracking and obstacle avoidance of the leader,formation control and formation reconfiguration.Then,a hierarchical structure was proposed for the problem of UAV formation control and reconfiguration,including,active disturbance rejection control(ADRC)based low-level control,virtual force based trajectory tracking and path following,liquid sphere-inspired formation control,and physicomimetics based reconfiguration strategies.A virtual force guidance law was proposed for trajectory tracking and path following of UAVs.Considering the kinematics model,virtual forces were designed to govern the vehicles.Using the guidance law,the UAV can track straight line,circle and general curve with time-varying curvature.When the reference trajectory is a straight line,the guidance law is equivalent to PD control.For circle or curve tracking,the guidance law is equivalent to feedback linearization method.By designing extra virtual repulsive force,the guidance law can be used to avoid obstacles on the reference trajectory.The use of artificial physics makes the guidance law have a large parameter adaptation and be computationally simple.Besides,the physical meanings of the parameters are definite,which makes it easy to tune in application.Simulation and experiment results demonstrated the effectiveness of the proposed guidance law.A liquid sphere-inspired physicomimetics approach was proposed for multi-agent formation control.The agents were formulated as a liquid sphere,which was modeled by a virtual spring network.Considering different formation structures,a liquid sphereinspired leader-follower approach and a liquid sphere-inspired virtual leader approach were proposed for formation control.The stability and convergence were proved.Using the proposed approach,arbitrarily shaped formations can be obtained by designing the control parameters.Comparing to the structure potential function approach,the required number of communication links is reduced using the liquid sphere-inspired virtual sphere approach.The use of physicomimetics makes the approach computationally simple,and the physical meanings of the parameters definite.These make the approaches easy to use in applications.Simulation and experiment results demonstrated the effectiveness of the liquid sphere-inspired formation control approach.A physicomimetics-based approach was proposed for the 3 types of reconfiguration problem.Inspired by some phenomenons of liquid,such as,merging and obstacle avoidance,we analyzed the scalability and flexibility of the liquid sphere-inspired formation control approach.Then a structure was proposed for the problem of UAV formation reconfiguration.Strategies were designed for the problem of reconfiguration due to mission update,UAV adding or lost,and obstacle avoidance,respectively.By redesigning the parameters,internal collision free switching between different formation shapes can be obtained.By local communication reconfiguration,an extra agent can be added to a formation like mixing a drop of liquid into a liquid sphere.By designing extra virtual repulsive forces from obstacles,a formation avoids obstacles like a fluid flowing over obstacles or squeezing through narrow passages.Simulation and experiment results demonstrated the effectiveness of the approach.A formation flight experiment system was developed.The experimental system included an X-Plane based hardware-in-the-loop(HIL)simulation subsystem and a flight experiment subsystem.In the two subsystems,the same autopilot and ground control station were used.By the HIL simulation,defects or bugs emerge in the HIL simulation can be found and modified before field experiment.Therefore,the time and cost required by the field experiment was reduced.Experiment results demonstrated that the autopilot validated by the HIL simulation can be moved onto the flight experiment subsystem directly.The only work required by the actual formation flight was to partially retune the control parameters.To reduce the cost of the research,HIL simulation was carried out to investigate the proposed approaches in the dissertation.The results showed that the proposed approaches are applicable in UAV formation flight.