342 / 1971-01-01 00:00:00

Specific Crossroad Recognition Using Sequential Information

Unmanned vehicles; Specific Crossroad recognition; Sequential recognition; Sigmoid function; Normalization; DPM; LSVM

Localization is quite important for unmanned vehicles. However, GPS cannot solve all the problems about localization. As crossroad is very important for unmanned vehicles, it is necessary for unmanned vehicles to recognize the crossroad and to load some other recognition tasks (such as traffic light recognition and traffic sign recognition) at the same time. The precision of GPS can hardly tell if the traffic lights or traffic signs are shown in images. Using images to help vehicles to localize themselves seems to be promising. Although some works have been done, the single-frame approach need to be improved. In this paper, two algorithms are provided. One is Bayes theory based sequential recognition and the other is normalization of model scores. The excellent performance of
these approaches is shown in the integrated experiments and illustrates this method to be promising in the future.