| Animal robot is an application which takes advantages of animal's natural abilities,combining with modern technologies as electronic,sensing and network control to realize the control of animals.This study,we chose craps as our participants with binocular stereo vision technology to study the control model of carps working in the maze task.We fully introduced the maze system and the control model.We applied electrostimulation technique to investigate the responses of cerebellum and medulla oblongata in two different conditions,which were the out-of-water condition and underwater condition.In the out-of-water condition,the carps were stimulated in different parameters by electricity at different brain regions underlying craniotomy,and we chose to analysis various coordinates.Then we chose these regions which were acting with significant fishtailing respond to be our target stimuli regions at under-water experiment.These carps were controlled with wireless control device in the maze,they accepted signal through the specific positions of cerebellum and medulla oblongata which were chosen as the positions with best response.The binocular stereo vision camera were set up with correct parameters.We used this device with frame difference and Gaussian Mixture model to analysis the carps' moving features,meanwhile we chose the skeleton extraction,particle extraction and trajectory reconstruction of videos features to analysis the carps movement characteristics.Curve fitting was applied to build carps movement model so that we could get the values of frequency and amplitude of carp tail.At last,we test the accuracy of displacement,speed,accelerated speed and rotation angle of coordinate and trajectories by reconstructing 3D trajectory,through the establishment of the BP neural network,the effect that the controllability degree of the cerebellum and medulla oblongata of the carp robot was evaluated,in this way,we analysis the feedback data to improve our control model. |
PDF Full Text Request