| Localization is one of the most important problems that an autonomous mobile robot need to solve. It is the key to robot navigation, simultaneous localization and mapping(SLAM) and other complex task. Localization itself is also a complex problem, which contains lots of fields in robotics, such as robot kinematics, environment representation, information fusion, uncertainty processing, etc. The localization of indoor mobile robot based on laser scanner is mainly discussed in this paper, here is the brief work:Mobile robot kinematics model is build. The principle and error model of two commonly use robot sensor are analyzed. The environment representation methods in robotics are also compared and analyzed, including geometry map, occupied grid map and topological map.The localization problem of mobile robot in indoor environment is analyzed and modeled. Two localize framework are analyzed and researched: Markov localization based on kalman filter and Monte Carlo localization based on particle filter. An improvement is make to Monte Carlo localization after an introduction of several scan matching algorithm for laser data association.Many experiments are designed and performed based on mobile robot to evaluate the localization effect in indoor environment, aiming at how initial distribution and particle number can affect the precision of localization.At last, this paper also introduced the SLAM problem and the relationship between localization and SLAM. The main work is extended into SLAM problem, for investigating the FastSLAM solution based on Rao-Blackwellized particle filter. |
PDF Full Text Request