CALIBRATION APPARATUS, CALIBRATION METHOD AND CALIBRATION PROGRAM

摘要

[Problem];To provide a calibration apparatus capable of reducing a workload for calibrating a position and a direction of each range sensor.;[Solving Means];In the measurement arithmetic unit 100, the social group identification portion 5610 identifies a group candidate that is detected as a group out of moving measuring objects in each laser range finder. The group comparison portion 5612 identifies a group in agreement for each pair of the laser range finder, and calculates the relative position of each pair of the laser range finder according to the position and the directions of the identified matching group. The network position identifying portion 5620 calibrates the position and the direction of each laser range finder in the sensor network coordinate system such that the error of the position of the object observed in common from each pair becomes minimum.

主权项

1. A calibration apparatus for performing calibration of positions and directions of a plurality of range sensors installed in an area, each of the plurality of range sensors measuring a distance between two points in a noncontact manner, comprising:
an interface device configured to receive measurement data from each of said plurality of range sensors; a storage device configured to store said measurement data from each said range sensor in a time series; and an arithmetic processing unit configured to perform said calibration based on said measurement data, wherein said arithmetic processing unit is configured to a) detect positions of moving objects in said area for each of said plurality of range sensors; b) identify a set of group candidates and pose invariant features of said group candidates, each said group candidate being identified as a group among said objects that are detected for each said range sensor; c) i) calculate a relative positional relationship between each pair of said range sensors by identifying a set of objects observed in common for each pair of said range sensors based on said group candidates and said pose invariant features of said group candidates, and ii) adjust said relative positional relationship so as to minimize positional errors of the objects observed in common for each pair of said range sensors; and d) perform calibration of the positions and directions of said plurality of range sensors in a network of said plurality of range sensors based on the positions of said objects observed in common with said minimized positional errors.