ファジー推論を用いたセンサ統合システム

抄録

This paper deals with a new integrated measurement system. This system consists of LED displacement sensors and CCD cameras. A spot-light of an LED displacement sensor is used for the active stereo sensing method. Simaltaneously, the outputs of the LED displacement sensors make it is easy to calibrate the camera parameters. Neural networks were employed to compensate such output errors of the vision system, as a camera parameter error, lens distortion and so on. Plural sensory outputs are integrated by a sensor integration system with two antecedents, one of which is the value determined by changes of the sensor output and the sensor position and the other one of which is the environmental data of the sensor. Since these information are vague values, the system is based on fuzzy inference processing. Furthermore in order to fast and accurate sensing, the system is based on an active sensig method, and the system only uses the vision system when the system estimates the LED displacement sensor cannot measure a distance. The effectiveness is verified by some experimental results.