The height of the object is determined by cameras and ultra
sensors. The distance to the obstacle is obtained by stereo vision.
If the correspondence between a pair of image feature point is
established, inverse projective-mapping holds. By this inverse
mapping, the geometry of obstacles in a three-dimensional space,
particularly the distance to the obstacle is obtained. Once the distance
is estimated, the height of obstacle is computed from calculating
the pixel image size displayed on screen. The estimated
value by vision has much error as the distance is far away from the obstacle, hence a fixed distance to the obstacle is always selected
before jump. The obstacle height H is determined as follows.