There are common problems with perception and avoidance technology, such as lagging bird recognition, insufficient accuracy in predicting disturbance behavior, and lack of refined quantitative basis for avoidance decisions. Therefore, this article constructs an efficient response framework for bird strike risks in low altitude scenarios, and introduces a collaborative mechanism between artificial intelligence and the Internet of Things to solve the core bottleneck of intelligent perception and control. This article achieves synchronous collection of key variables such as bird density, relative velocity, and invasion angle through real-time interconnection between IoT multi-source nodes and UAV onboard sensors. At the same time, AI behavior modeling is used to dynamically predict the escape trend, approach probability, and disturbance sensitivity of bird flocks. In terms of control strategy, based on the risk gradient output by the behavior model, the maneuver avoidance path with the minimum trajectory deviation is automatically generated, and the triggering timing and threshold of sound, light, and airflow are finely controlled to balance safety and energy consumption constraints in the avoidance process. The experimental results show that the method performs well in detailed indicators such as bird situation recognition accuracy, disturbance prediction error, trajectory exposure, and avoidance energy consumption: in the four test routes, the trajectory exposure is 1.403, 0.240, 2.831, and 1.122 (arb. nits), respectively; avoiding energy consumption is 17.4%, 6.0%, 29.3%, and 17.2% respectively; the number of drone avoidance actions is 4, 2, 2, and 4 times. The proposed mechanism can finely control the trajectory of unmanned aerial vehicles in actual low altitude flight segments, while achieving efficient risk response.
Liu et al. (Thu,) studied this question.