One of the most important aspects in the navigation of ALV is computer vision or machine vision. Usually, it is achieved by using multisensor fusion technology. As we know, laser radar is a typical sensor in this project, especially in the situation that there is an obstacle in a road. It is often effective to describe the relationship between road and obstacle by using height matrix from range data. However, when the front view is more complicated, such as a wall or a building on which exists a hole or a corridor big enough for ALV to go through, the above method may not be well done. For this reason, we propose a novel approach by using two matrixes from the range data to solve the problem. The main idea is that from the range data we figure out two matrixes, one is the height matrix, representing the height of the object from the horizontal plane, the other is the depth matrix representing the depth of the object from the laser radar vertical plane. By using the information of both height and depth, we can understand the front environment more precise and better.
|