Research On Position Estimation System Of Micro Quadrotor

In recent year, the micro quadrotor has caused more and more attentions of researchers in the field of robotics, aviation and automation. Among, the position estimation is very important. At present, most existing approaches to indoor localization focus on using either cameras or laser scanners as the primary sensor for pose estimation. Although laser sensor can achieve the high-precision distance information and the noise is small, finding scan point correspondences across scans is challenging as individual scan points lack unique attributes, such as corridor, hall and so on. The method that using camera can get the colour and texture informations of environment, and can find corresponding points in consecutive images, but one has to deal with images with few or no visual features as well as consider scale factor ambiguities to recover absolute distances.Aiming at this problem, this paper develop the laser scan augmented visual odometry algorithm to realize the autonomous positioning of a micro quadrotor indoors, where GPS signals are non-existent, laid a technical foundation for micro quadrotor’s practical application. The algorithm in this paper can alleviate the drawbacks of each individual modality by fusing a camera and a laser scanner, and then improve location accuracy and stability. The main contents are as follows:Firstly, introduce linear camera model, explain the relation between pixel coordinate and world coordinate, then introduce the sensor calibration methods, mainly include two major parts:one is camera’s intrinsic parameters, the intrinsic parameters can get by using Zhang Zhengyou’s calibration principle, the other is the extrinsic parameters between the camera and the laser sensor, on the basis of the result of the intrinsic parameters, it is easy to obtain the extrinsic parameters, and then get the position relationship between the camera and the laser sensor. These works make good preparations for the subsequent laser scan augmented visual odometry algorithm.Secondly, introduce the whole diagram of the algorithm, analyses every step in detail, the study mainly include several algorithms as follows:SIFT, RANSAC,8-Point algorithm, Triangulate, and so on. Get the result from these algorithms, based on these results, get the micro quadrotor’s position information.Finally, based on these theoretical studies and the data from the sensor, the system is studied by simulation using Matlab. The simulation results show that the error can converges to the zero point and gain a good result of position estimation.