Özet:

This paper presents a feature based docking strategy with a LiDAR for a non holonomic mobile platform to be charged at the charging station. The docking station has only a V-shape structure to be detected from the LiDAR scans. The proposed docking algorithm consists of three main sections which are the localization of the V-Shaped feature with the LiDAR measurements, the path generation with a quintic polynomial and the Proportional Integral Derivative (PID) implementation to track the generated points of the path and correct the yaw angle along the way. Once the relative position of the docking station is detected by matching the synthetically generated V-shape feature with the current Lidar scans, the trajectories are generated by calculating the coeeficients of the quintic polynomial. Afterwards, the initial states, the states and the yaw angles at the points along the path are taken as inputs and the algorithm corrects the robot’s yaw angle as it follows the path by a PID control. The algorithm is both tested within the simulation environment and the real test environment. The platform docks with a good accuracy which is around ± 2 cm in x and y, ± 1o in . With the proposed strategy, additional sensors are not necessary on the platform or on the charging station. Moreover, the robot can navigate even in the absence of light with a good accuracy. Hence, a cheaper solution is obtained for an important problem of autonomous mobile robotic applications.

Anahtar Kelimeler: