Design And Research Of Audio Visual Fusion Guide Robot

Artificial intelligence technology has gained rapid development in today's society.As an important field of artificial intelligence technology,intelligent robot has attracted people's attention and is entering various fields of national life.Especially in recent years,the large-scale industrialization of commercial and domestic robots,service robots are gradually becoming a new field of robot technology promotion.However,the blind number increase,because the blind can't get information by means of vision,only through other senses to obtain information,a lot of inconvenience to the daily life,so the design of a safe travel service can guide the blind type guide robot has a very important significance.Aiming at the problem of blind people travel,a set of audio-visual fusion guide robot system is designed.The main research work includes the following aspects:1.According to the defects of the blind innate physiological,and a combination of blind daily activities,analyzes the system structure and function of the circuit;and then puts forward the hardware structure diagram of guide robot system,and separately from the control part and detection,information transmission part and a driving part four to guide robot function and the hardware circuit of.The hardware system of the guide robot is completed.2.The embedded development environment of the guide robot is constructed,and the software flow of the robot system is designed.Then,the multi-sensor information fusion algorithm based on BP neural network,path planning based on artificial potential field,GPS remote location algorithm and human-computer interaction function are studied respectively.The software system design of the guide robot is completed3.Respectively on the recognition algorithm of the zebra crossing,traffic lights blind and three typical traffic signs in-depth study.The zebra crossing recognition,respectively from the bipolar coefficient,the uneven image block thought,mathematical morphology,threshold segmentation and edge extraction of the four aspects of research;for blind identification,respectively,from the extraction of color feature image threshold segmentation and morphological processing and edge extraction are studied in three aspects;for traffic lights recognition the traditional method,and the combination of deep learning,put forward a point spread target detection algorithm.Three aspects of CNN structure design,CNN model training and random point spread algorithm are studied respectively.It solves the problem of low recognition rate and poor robustness to complex environment in traffic sign recognition.4.Simulating the working environment of the robot,testing and analyzing the designed guide robot under the actual road conditions.The experimental contents include obstacle identification test,traffic sign recognition test,remote location test and man-machine cooperation test.The experimental results show that the guide robot has the advantages of high recognition accuracy,robustness to complex environment and so on,which meets the needs of blind people.