Needs of the autonomous vehicle in indoor environment such as factory storehouses are increasing. Variety types of indoor vehicles are available in commercial market. However, in practice to realize the autonomous vehicle, it is required to change the whole arrangement of storehouse, because of the lack of intelligence of the vehicle. Our aim in this paper is to make autonomous vehicle more intelligent. The geometric information of the factory storehouse changes frequently due to the turnover of the cargo. To let the vehicle run safely, real time map construction is necessary. Here, we apply in such a situation the sensor fusion technique by use of the Omni directional image sensor and laser radar. We propose a new update algorithm called SMT (Shift Matching Transform) that obtains region segments with trusted semi-real-time geometric map. The key point of proposed update algorithm is the fusion of laser radar and Omni directional camera. These have same property that keeps profile of angles. By using developed the new algorithm SMT, the processing time was drastically reduced. Successful experimental result was presented at processing interval is less than 500 msec. We demonstrate to find arbitrary complex walls to determine the absolute position of the vehicle. Based on experimental results on the corridor, we demonstrate semi-real-time map reconstruction.
展开▼
