Figure 3.
Simulation, investigating drift-compensating abilities given varying predictive position estimation errors. Comparison of position estimation error at the end of the trajectory between Algorithm 1 and a particle filter for localization in a known map with varying predictive position errors at the initialisation of the simulation. The lines connect the average results after 100 Monte Carlo repetitions with different realisations of the odometry noise, and the error bars represent one standard deviation. (a) Estimation accuracies with varying predictive position error. (b) Estimation accuracies with varying length scales .