A robot pose estimation approach based on key feature registration

Pose estimation is a fundamental issue in the research of mobile robot. A robot pose estimation approach based on key feature registration is proposed in this paper. The key features in the environment are extracted firstly. When dealing with many non-corresponding points, registration has a slow convergence rate, and gains poor performance. To solve these problems, a Key Feature Based Accelerated Iterative Closest Point (KF-AICP) algorithm is presented for key feature registration. Then the current robot pose can be estimated based on the result of registration. The dataset is established using 2D laser radar in indoor scene. The experimental results show that the proposed approach achieves good accuracy and real-time in pose estimation of mobile robot. The validity of proposed approach is verified.