Research On Hybrid Navigation System Of Indoor Robot

In order to solve the problems of unstable light source and insufficient reliability of fully autonomous navigation in low-cost indoor navigation robot,this paper proposes a low-cost indoor navigation robot scheme based on visual guidance for multi-navigation technology fusion.The scheme includes the manual calibration of the driving area under the visual guidance of the fish-eye camera,and the reference database positioning mode based on the mobile phone's built-in sensor and camera,which mainly completes the following aspects of the work.Firstly,a method of obtaining location data by means of a mobile phone built-in sensor is proposed and the feasibility of the method is verified.According to the sampling period to obtain the data of the acceleration and other sensors in the mobile phone,the positioning parameters such as the speed and displacement of the robot are calculated by mathematical operation and compared with the real displacement values.Secondly,the collaborative controller for indoor navigation robot is designed and machined.It is composed of a host controller based on a mobile device and slave Controller Based on Microprocessor.The controller includes system power supply,communication module,motor drive module,feedback module,ranging and obstacle avoidance system,etc.Thirdly,the scheme of calibrating the driving area of indoor robot by fish-eye camera is proposed.The fish-eye camera on the roof gets the indoor area photo,and the fish-eye correction algorithm completes the correction of the picture.In combination with the size information of some known objects in the room,the coordinates of the key objects in the room are obtained,thus the feasible region of the robot can be calibrated.Finally,the establishment of projection feature database and the comparison and analysis of image feature matching algorithm are realized.The phone's camera gets photos of the roof reference and processes the image.The image data in the database corresponds to the indoor coordinates.A variety of image matching algorithms are adopted to match the photos and analyze the matching results in order to obtain the current coordinates of the indoor robot under visual navigation.The computational results show that the algorithm has some limitations,but it is more acceptable than the inertial navigation results.