Abstract: In order to meet the demands of obstacle avoidance and path planning for micro aerial vehicle (MAV) in an indoor environment,we establish a low cost embedded binocular stereo vision platform based on a BeagleBoard-xM board and consumer-grade cameras. With the indoor environment information obtained through the binocular stereo vision system,based on a 3D space description model-octree and an inverse sensor model and combined with the attitude information of the MAV,we propose a 3D map-building method described with a 3D occupancy map. The result of the experiment shows that the 3D occupancy map acquired by embedded binocular stereo vision system described the indoor environment of MAV accurately and effectively. It can thus be used widely in unmanned aerial vehicle navigation in an indoor environment.