Simulation Research On Visual Autonomous Obstacle Avoidance Technology Of EVTOL Aircraft

As an electric vertical takeoff and landing vehicle,the eVTOL has the characteristics of cargo and manned flight.With the development of science technology and economy,eVTOLs will be operated commonly in urban low altitude environment,which has high requirements for the stability and safety of the aircraft,especially for its function of autonomous obstacle avoidance.The addition of the autonomous obstacle avoidance design can not only effectively ensure flight safety and improve the work efficiency of the remote captain,but also in the special flight environment where the eVTOL remote operator cannot operate the aircraft in time or the control signal is interfered,the obstacle avoidance system will effectively reduce the accident rate of aircraft damage and injuries to persons and structures caused by operation errors.The obstacle avoidance technology of eVTOL aircraft in the urban low-altitude environment was researched in this thesis.An autonomous obstacle avoidance scheme based on vision was proposed by integrating machine vision and obstacle avoidance algorithm,and the feasibility of the scheme was demonstrated through simulation verification.Firstly,the dynamic mathematical model of eVTOL was built based on the principle of the rigid body motion model and the dynamic principles such the transformation relationship between coordinates,tension and torque in this thesis.Visual information was collected using the depth camera combined with the SLAM algorithm framework.The feature extraction and matching were completed by the feature point method.The aircraft pose was calculated by the ICP method.The dictionary was trained on basis of the bag of words,and the image similarity was estimated,and the cumulative error problem of visual positioning was solved through detection loopback,and the positioning reliability was improved.Then,the principle of A-Star algorithm was analyzed,and the obstacle avoidance effect was verified in two-dimensional and three-dimensional environments respectively.An EGOPlanner algorithm based on A-Star algorithm was applied to better meet the stability and reliability requirements of eVTOL.The smoothness of path planning results was assured by introducing the polynomial relationship of collision penalty,combined with B-spline curve optimization.The real-time performance of the algorithm was verified by simulation experiments.Finally,the evaluation and analysis of the simulated flight parameters verified the stability of the obstacle avoidance process under the ROS platform of the simulation experiment environment.The corresponding urban environment layout was completed in Gazebo 3D physical simulation software according to the real situation of urban high-rise buildings.Obstacle avoidance results were intuitively displayed combined with the rviz 3D visualization tool.Multiple groups of random tests were used to verify the success rate of obstacle avoidance,which ensured the feasibility of the visual autonomous obstacle avoidance technology scheme studied in this thesis.It provided a new idea and framework system for the subsequent research and design of eVTOL.