The locust’s lobula giant movement detector(LGMD)neurons shows a preferential respond to objects close to the locust’s field of view during direct collision

but the lack of motion direction sensing mechanisms in this network leads to the lack of omnidirectional obstacle avoidance capabilities in the locust’s visual neural network.In order to improve the omnidirectional obstacle avoidance ability of the locust collision perception network

inspired by the fruit fly visual neurons

the artificial locust visual neural network LGMD is combined with the elementary motion detector(EMD)

and the innovative moving target extraction algorithm is fused to reduce the network complexity.A bionic vision obstacle avoidance neural network(BVOANN) and an applied motion control strategy for dynamic collision detection are proposed.The Visual Studio software is used to simulate and analyze the video obtained from the real world

and verified on the low-cost robot.The results show that compared with the traditional model

which can only detect radially moving objects

the improved model has all-round moving object detection function

so as to achieve better obstacle avoidance effect.