Learning Behavior Control Algorithm for Micro Flapping-Wing Aerial Robot Based on Clustering Data-Driven

In order to solve the problem of poor adaptability and generalization ability of traditional control algorithms for Micro Flapping-Wing Aerial Robots (MFARs) and the limited application of traditional control algorithms, the use of task-oriented and job-object-oriented data training and data-driven approaches for aerial robots is utilized to achieve navigation control of MFARs based on machine vision. This paper proposes a learning behavior control algorithm for MFARs based on cluster data-driven. The data from the gyroscope of the flapping-wing machine is processed using cluster analysis to obtain the directional characteristic behavior function of the flapping-wing machine. Cluster analysis is utilized to process the visual images of the Flapping-Wing Aerial Robot (FWAR) in order to obtain the image area response behavior function of the aerial robot. This enhances the generalization ability of the behavior control function of the FWAR with perception features. While ensuring the FWAR’s vision-based navigation, the features of the FWAR’s visual target and environment are extracted through cluster analysis, providing technical support for the behavior control and target tracking of the FWAR based on dynamic vision. An experimental verification system was built using a FWAR system and an onboard camera. Experimental verification was conducted on the learning behavior control algorithm for FWARs based on cluster data-driven. Three verification experiments were conducted to demonstrate the effectiveness of the learning behavior control algorithm.