We have estimated the position and attitude of
mobile robots using particle filters on the basis of the
likelihood of their positions, which were calculated by
matching LRF measurement results to 3D point cloud data
measured using MMS. We have confirmed through run test
comparisons in outdoor environments with carrier-phase
GPS/INS navigation that we can determine the position of
mobile robots within 30 cm without using GPS and we can
reduce the errors due to sloping road surfaces or DR with 3D
voxel map.