Research On Bionic Optical Flow Navigation Technology And Its Application In Multi-aircraft Flight Missions

Numerous biological studies have shown that insects have a very simple visual system with very limited ability to extract and process data,yet can accomplish complex behavioral control such as flight and landing using only visual information.Insects rely on optical flow.Specialized neurons of tangential cells respond to wide-field patterns of retinal motion.They resolve complex optical flow patterns over large fields of view to extract navigational information to perform various tasks.Therefore,studying the principles of visual navigation in insects can provide an innovative reference for autonomous navigation of micro air vehicles.Based on the traditional optical flow method and the Wide-Field Integration method of optical flow,this paper perceives the spatial relative relationship between obstacles around the aircraft and the body in real time,and provides basic navigation for aircraft swarms in typical scenarios.information.The research work of this paper mainly includes the following aspects:1.A speed estimation method based on optical flow/inertial fusion is proposed.This method does not require prior knowledge.It perceives the environment through visual sensors,fuses Inertial Measurement Unit and altitude sensors,and connects metric velocity with the point correspondences between successive images.To improve estimation accuracy,a Mean Shift（MS）based method is used to detect outliers and select the best matching points.Combined with visual motion constraints and a vector dynamics model,the velocity is estimated using a standard linear Kalman filter.Finally,the proposed method is validated.2.Based on the parameterization of a simple three-dimensional environment,an algebraic model of the optical flow WFI is established,integrated over a wide optical flow region,and using the weighted sum of the instantaneous modes of optical flow,the proximity and velocity estimates associated with these environments can be extracted,which makes motion estimation robust to unknown environments as well.Verify the effect of the camera optical axis direction on the WFI estimation accuracy.At the same time,because the optical flow patterns observed in a limited field of view can be quite similar for different motion patterns.Therefore,considering the WFI state estimation with multiple camera sensors,it is verified that the estimation accuracy is effectively improved as the number of sensors increases.3.A state estimation system based on optical flow WFI is constructed,and the obstacle perception method based on optical flow is added on this basis.And the experimental verification and analysis were carried out in the multi-UAV enclosed collaborative scene.