Application of Particle Filtering Technique for sensor fusion in mobile robotics

Accurate position estimation is very essential for successful operation of any autonomous mobile robots. Generally, Extended Kalman Filter (EKF) is used to fuse multiple low cost sensor information for better position estimation of mobile robots. However, due to the first-order approximation while performing linearization of the nonlinear model in the EKF, it will introduce large estimation errors over the time. In order to reduce the significant estimation errors, the Particle Filter (PF) is presently used as a modern sensor fusion methodology applied to mobile robotics due to its generic nature to tackle uncertainty & nonlinearity. This paper illustrates application of Particle Filtering Technique for reliable estimation of the state vector of a mobile robot in association with proprioceptive (Odometry) and exteroceptive (Laser Range Finder) sensors for efficient control. The paper also presents a comparison of the performance of the EKF & PF techniques for the estimation of the states & control of the mobile robot and establishes the superiority of PF over EKF.