抄録
We propose a strategy for performing under the environment with movable obstacles. Movable obstacles are detected by using active sensing and the color range sensor. Active sensing is that the robot tries to push an object using its arm and classify them into movable or non-movable object based on fused results from the force sensor, the joint angles, the color range sensor. The color-range sensor consists of TOF (Time Of Flight) range sensor and conventional color stereo cameras. The proposed strategy is that the robot continues to detect the movable obstacle, move the obstacle and reconstruct the environment recognition. Finally, we show the experimental result that the humanoid robot HRP-2 can navigate in the environment with movable obstacles by means of the proposed strategy.
