Method and system for autonomous flight control of unmanned aerial vehicles

Abstract

The invention discloses a UAV autonomous flight control method and system. The invention first creates a generation network and a classification and discrimination network; determines a false sample according to the training set A and the generation network; then combines the false sample with the The training set A is input into the classification and discrimination network for training, and the output parameters of the classification and discrimination network are obtained; the network output result has one more dimension than the scene instruction label vector, and this dimension feature is used to classify true and false samples, and the training generates the network. In the process, the fake samples without instruction marks and the training set A with instruction marks can be used as real samples and input into the generation network to obtain the output parameters of the classification and discrimination network, thereby improving the accuracy of determining the parameters of the classification and discrimination network, and further improving Autonomous flight control precision.

Description

technical field [0001] The invention relates to the technical field of UAV flight control, in particular to a UAV autonomous flight control method and system. Background technique [0002] With the continuous development of transmission lines, the demand for inspections of transmission lines continues to increase. The addition of UAV inspections has greatly improved the efficiency of power inspections. However, the current operating mode of UAV inspections still remains at the In the manual operation stage of patrol personnel, the technical threshold required for operators is relatively high. Realizing the autonomous flight of drones can greatly reduce the labor of inspection personnel and facilitate the promotion of drone power inspections. [0003] The existing autonomous flight control of UAV mainly includes the following solutions: [0004] (1) Based on the lidar to detect obstacles, realize the perception of the surrounding environment to achieve the purpose of UAV ob...

AI Technical Summary

Problems solved by technology

[0004] (1) Based on the lidar to detect obstacles, realize the perception of the surrounding environment to achieve the purpose of UAV obstacle avoidance, and realize the autonomous flight control of UAVs in specific working scenarios; its advantages lie in high precision and long distance , but at the same time, the lidar is more sensitive to the environment, and it is easily interfered by strong light in the outdoor environment, which not only affects the flight safety of the UAV, but also affects the control accuracy of the autonomous flight.

[0005] (2) The UAV obstacle avoidance technology based on the ultrasonic module, using the ultrasonic module to detect the obstacles around the UAV, combined with the obstacle avoidance algorithm to realize the autonomous flight of the UAV; its advantage is that the hardware cost is low, and the processor The computing power requirements of the drone are not high, but because it uses an ultrasonic module, it has a strong dependence on the temperature of the environment, and the obstacle recognition efficiency is low, which not only cannot meet the requirements of high-speed flight of outdoor UAVs, but also affects the accuracy of autonomous flight control.

Images