| Summary: | 碩士 === 國立虎尾科技大學 === 資訊工程系碩士班 === 106 === This work combines the sensory data of odometers and a Laser Range Finder (LRF) through the Particle Filter (PF) algorithm to realize an indoor positioning system on a wheeled robot. The system uses a particle set and the odometers to predict the posture of the robot, and then the data points measured by the LRF are clustered by an Adaptive Breakpoint Detector (ABD) algorithm. An Iterative End-Point Fit (IEPF) algorithm is then performing line segment fitting on the clustered laser points to generate a local map. Each particle will be assigned a weight according to the comparison results of this local map and a built-in environment map. Finally, a resampling procedure is performed in accordance with each particle’s weight to generate a new particle set to correct the estimated posture of the robot. This work uses a Field Programmable Gate Array (FPGA) based platform as the hardware control platform, which is responsible for data acquisition of motor encoders and motors control. A remote computation platform is used for receiving motor encoder data from the hardware control platform and measurement data of the LRF. After a series of computation, the remote computation platform transmits the final control command back to the hardware control platform to drive the motors. Simulation and experimental results show that the resultant system can accurately correct posture errors of the robot within ranges, where the position error range is ± 0.1m and the orientation error range is ± 5 degree.
|
|---|
