Reactive Collision Avoidance of UAVs withStereovision Sensing.

摘要

Using simple passive stereovision pin-hole sensors, an effective reactive collision avoidance algorithm is presented in this report for unmanned aerial vehicle (UAV). Both kinematic model as well as point mass models of UAV are considered for system dynamics. The stereovision camera is considered mounted on the UAV. The camera sensors and the obstacle are forming a geometrical configuration and triangulation methods are used to estimate the position and velocity of the obstacle falling under the field of view of the camera sensors. For simplicity, shape of the obstacle is considered as spherical. The position estimation of the obstacle is also carried out by image processing methods wherein the feature points detection, stereo matching and triangulation algorithm are used to compute the 3D reconstruction of obstacle. This method uses the computer vision system toolbox and VRML viewer of MATLAB(registered trademark), Using this we are able to compute the approximate size and centre of the obstacle. This geometrical formulation takes advantage of both 'stereo vision' as well as 'optical flow' signatures and hence it is capable of estimating the range information as well, making its position estimate quite accurate. A large number of simulation studies, which also includes consistency checks for filtering algorithms, leads to the conclusion that this strategy is quite effective in avoiding popup obstacles within a very short time and hence can be very useful for reactive collision avoidance.