Abstract
This technology is intended to obtain cross-sectional shapes and positional information of underground pipes which cannot be targeted as are open spaces on the ground (in case of spaces where GPS coordinates cannot continuously be acquired). For cross-sectional shapes, point cloud data is acquired by a laser scanner in a transverse direction relative to an object. For positional information, the self-location and position of the robot are estimated by odometric data collected via IMU, and the estimate value is corrected via shape matching by ICP (Iterative Closest Point) algorithms using the laser scanner in longitudinal and horizontal directions relative to an object. To improve accuracy, the correct coordinate value is given to the robot at the starting and end points, and correction of whole measuring section is conducted in coordination with GraphSLAM.
