No abstract

This paper presents a development of guide-dog robot system for visually impaired. Based on a hall-sensor joystick and ultrasonic sensors, a "smart rope" system is designed for the human-robot interaction. In this system, multiple functions are provided for the self-walking in urban system, such as following, navigation and obstacle avoidance. To distinguish between small involuntary force and the intended navigational movement, a fuzzy logic control method is applied to improve the accuracy for the "smart rope" system manipulation. To compensate the lack of visual sense of visually impaired, a smart phone with camera is utilized as the robot vision, in order to detect the traffic lights and the zebra crossing. A fast vision recognition approach is provided based on Adaboosting and Template matching combined algorithm. For the evaluation of proposed method, an integrated system is implemented to the mobile robot platform. The performance of both interactive system and vision system are analyzed after the experiment in the urban environment. System's accuracy, usefulness and adaptability are verified. The experimental results showed that this new designed guide-dog robot system is suitable and effective to assist the visually impaired for the self-walking.

No abstract

This paper presents a research about a vision of guide-�og robot syste � for vi � uall � impaired �eople via the camera in smart phone. In this system, camera can replace bhnd people or visually I � palred � eople s eyes to analyze the traffic situation. The recognition of traffic lights is pr � posed through the adapt � ve bo � stmg (A . d � boost) algorithm and template matching algorithm combined approach. Firstly, Adaboos� method .IS . app�led for tramm� the Pedestrian light signal detector and secondly by applying template matching algonthm t � dlstmgu � sh the . traffic hghts between green and red. And vanishing point is applied for zebra-crossing detection, and histogram IS apphed to defend the misrecognition and separate real zebra-crossing from staircase.

Path planning problem is the major issue of the control of mobile robot. Among the path planning methods the artificial potential field theory is widely used for mobile robots, because it provides simple and effective motion control input. But sometimes it is hard to detect exact obstacle shape, because some obstacle can move itself, thus we can detect its probability only. So we suggest the path planning method using potential field with the probability information. After that we simulate our algorithm and show the results.