Simultaneous Localization and Mapping (SLAM) is receiving much attention in robotics and control community in an effort to build autonomous actuators, which can survive in uncertain and unexplored environment. FastSLAM, introduced by Montemerlo is an efficient and robust solution in the field of localization. The core idea of FastSLAM revolves around the Rao-blackwellized state, where the trajectory is represented by weighted samples and the map is computed analytically. This approach uses a particle filter in which each particle carries an individual map of the environment. Accordingly, a key question is how to reduce number of particles for accurately map the environment where the motion noise is very high. The paper illustrates two approaches FastSLAM1.0 and FastSLAM2.0, which differ by their sampling strategy. In general the performance of FastSLAM2.0 is similar to FastSLAM1.0 but it varies particularly in situations in which the motion noise is high in relation to the measurement noise. The comparison of their accuracy is shown by the simulation results for different number of particles and high motion noise.
Robotics, SLAM, Particle Filter, RBPF, FastSLAM2.0
Manjula Sahu; Chetan Kamble, Comparison of FastSLAM1.0 and FastSLAM2.0 for Relatively High Motion Noise Environment, HCTL Open International Journal of Technology Innovations and Research (IJTIR), Volume 14, April 2015, eISSN: 2321-1814, ISBN (Print): 978-1-62951-946-3.