| Level of robotics development has become an important symbol of national high-tech strength and prosperity. With in-depth research on interdisciplinary, robots are no longer merely used for simple mechanized operations, the research gradually transited to the purpose of improving robot's autonomy.Mobile robot localization is the basis of autonomous robot research. To solve the problems of lacking prior knowledge and unexpected uncertainty effect in unknown environment, this paper studies on mobile robot localization mainly from the following aspects:An experiment platform for mobile robot localization was established: the platform hardware consists of IRONC robot, laptop and laser range finder. By object-oriented programming, system software achieved communication between PC and mobile robot based on EtherCAT, and built perception system for mobile robot based on LRF. In addition, kinematics analysis for non-holonomic wheeled mobile robot was executed and coordinate system of mobile robot localization process was established.An environment matching based relative localization method for mobile ro-bot in unknown environment was proposed. Firstly, the uncertainty in distance measurement of sensor data was analyzed based on modeling the localization problem, then by defining error model in matching process, a matching error sample was established with sensor data from statistics view. Finally, the relative pose of mobile robot was obtained through optimizing the sample based on Maxi-mum Likelihood Estimation.The presented mobile robot localization algorithm was experimentally eva-luated and validated in terms of effectiveness and robustness on the IRONC plat-form. The obtained results suggest that algorithm has advantages in accuracy and convergence speed, as well as the strong robustness in the case of perturbed ini-tial pose guess and partial non-static environment, which ensure the localization algorithm are effective in certain dynamic unknown environment. |
PDF Full Text Request