| Summary: | 碩士 === 國立臺北科技大學 === 機電整合研究所 === 88 === A behavior fusion strategy for automatic guided vehicle based on neural fuzzy networks is proposed in this thesis. In order for navigating in the unknown environment, the environment is classified into twelve typical environment modules. Five response behaviors with respect to these twelve typical environment modules are designed including image targeting, obstacle avoidance, narrow alley navigation, wall following as well as back off moving. Fuzzy neural networks are designed to infer the encountered environment module based on the sensor information. With respect to the inferred environment modules, the response behaviors are integrated smoothly so that the automatic guided vehicle is able to navigate in the unknown environment and reach the preset target. It will be shown in this thesis that the inference accuracy is increased with the proposed behavior fusion strategy. Moreover, the amount of training data for the fuzzy neural networks can be greatly decreased. Different experiments are made showing that the proposed strategy has satisfactory performance.
|
|---|
