Design Of Control And Decision-making Method For Low-altitude Integrated Operation Collision Avoidance System Based On MDP

In the future,it has become a development trend for medium and large UAVs to enter low altitude airspace and operate with man-machine integration.However,how to ensure the safety of aircraft operation in low altitude integrated airspace is the focus of current research.The airborne collision avoidance system provides the traffic conditions around the aircraft and gives maneuvering suggestions for different traffic conditions for aircraft pilots.However,the airborne collision avoidance system is widely used in large transport aircraft and civil aircraft,and it cannot be fully applied to the future low-altitude integrated operation scenario.The pilot performs actions according to maneuver suggestions to avoid collisions with other aircraft in the air,and plays a key role in the process of dealing with the decision and recommendation issued by the airborne collision avoidance system.However,the time delay in the process will bring uncertainty and instability to the control system.This thesis studies the problems existing in these two parts,and the main research contents are as follows:Firstly,the pilot was divided into MAV and UAV based on the integrated operation scenario,and several MAV/UAV pilot models were introduced.The simulation experimental models of human in the loop were built from three aspects with time delay effect:communication link,control response and rate limit,and the difference of communication loop between MAV piloting and UAV piloting in the integrated operation scenario was compared.The influence of dynamic characteristic response rate of pilot’s joystick force sensing response model on human-machine system was obtained.The effect of rate limiting on the output of human-machine system was analyzed.After that,the collision avoidance logic of integrated operation in the vertical direction was designed.Based on Markov Decision Process(MDP),the aircraft encounter process was modeled,parameter variables required by the MDP model were determined.And the MDP model was solved by dynamic programming method.According to the current decision consulting state,the action that can be selected at the next moment can be determined,and the corresponding acceleration can be selected,and the state transition probability can be obtained by sampling and interpolation methods.Through the expected cost function,the generation value of each state point after taking each action is obtained,and the generation value is stored as the collision avoidance logic table,where the action corresponding to the minimum generation value was the optimal maneuver mode to be selected.Finally,the relationship between pilot and collision avoidance logic was studied in this thesis.The applicability of the collision avoidance logic algorithm was verified by designing four simulated encounter scenarios between aircraft,and the pilot response delay in the collision avoidance scenario is discussed.The time when the pilot receives the instruction issued by the airborne collision avoidance system was postponed,the impact effect of the collision avoidance logic was studied in the case of time delay.The simulation results show that the collision avoidance algorithm designed in this thesis was also applicable when there is time delay.