Interaction with In-Vehicle Electronic Systems: A Complete Description of a Neural Network Approach

摘要

Interaction with in-vehicle systems (car phones, traffic information, route guidance, etc) becomes a very difficult task since the control devices are often reduced to small switches and push-buttons. To solve the problem, a new input interface is proposed, based on character recognition. The paper describes in detail how a simple neural network can be applied down to the level of an industrial realization to provide a reliable user-machine interface. The industrial application is that of character recognition where characters are drawn with the finger on a small touchpad. Compared with the nearest neighbour method, the neural network solution has slightly better recognition rate, is faster and requires less memory space. The design of the recognition system is given and results of an experiment made on a driving simulator are presented.