Cognitive vision for driving environment categorization using network-type fusion

For the next-generation automotive technology in advanced driver assistance systems, it is important that automobile itself understands driving environments autonomously. Therefore, we have been examining a type of cognitive vision system which categorizes the driving environments using an on-board vision system. Our approach is not the environment categorization by the complicated image processing, but by the network state from simple image processing modules. We modularize some simple image processing methods and make a fusion network by mutual evaluation between these modules from immune network point of view. Then, we associate the network state and the driving environment. In this paper, we show the validity of this proposed method by interpretation of a preceding vehicle lane change as an example of an experiment.