Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
4948738 | Robotics and Autonomous Systems | 2017 | 14 Pages |
Abstract
We propose a novel calibration method for industrial Automated Guided Vehicles (AGVs) adopting the tricycle wheeled robot model and equipped with an on-board exteroceptive sensor. The method simultaneously estimates the calibration parameters for the odometry and the exteroceptive sensor using only the input commands and the sensor egomotion of the robot while executing segment paths. Two AGV models, both relevant to industrial practice, are considered: the standard tricycle model and an asymmetric one that takes into account the different weight distribution in forward and backward motions typical of industrial AGVs. The parameters of the standard model comprise the steering offset and driving scale, which measure the angular offset of the tricycle steering wheel and the distance increment corresponding to an encoder tick, and the three parameters representing the sensor pose. The asymmetric model adopts different values for the steering offset in forward and backward motions to account for the different weight distribution. Closed-form or compact solutions are provided for both problem formulations. The observability of the calibration procedure is also formally proved. The proposed automated calibration procedure has been implemented on industrial AGVs, leading to estimation of the parameters in about 12 min, a significant improvement compared with one hour or more required by manual AGV calibration. Experiments with AGVs of various sizes in warehouses have assessed the effectiveness and numerical stability of the proposed approach. The precision of calibration parameters has been found to be about 0.1â for angles and 6Â mm for positions. Parameters obtained via the proposed automated calibration procedure have allowed different AGVs to accurately stop at the desired operation points.
Keywords
Related Topics
Physical Sciences and Engineering
Computer Science
Artificial Intelligence
Authors
Fabjan Kallasi, Dario Lodi Rizzini, Fabio Oleari, Massimiliano Magnani, Stefano Caselli,