Principle of Bayesian Robot Localization
Robot Localization as State Estimation (1) Lt: position of the robot at time t Given: Map and sensor model: Motion model: Initial state of the robot: Data Sensor information (sonar, laser range-finder, camera) oi Odometry information ai Wanted:
Motion Model Translational and rotational error are normally distributed represented by independent distributions
Robot Localization as State Estimation (2) Markov! Markov! Motion: Perception: … is optimal under the Markov assumption Kalman filters, Hidden Markov Models, DBN
Grid-based Markov Localization
Probability of a Laser Scan
Localization with Laser Range-finder