Triangulation-based fusion of sonar data with application in robot pose tracking

Publisher Website
Google Scholar
Abstract
In this paper a sensor fusion scheme, called triangu- lation-based fusion (TBF) of sonar data, is presented. This algo- rithm delivers stable natural point landmarks, which appear in practically all indoor environments, i.e., vertical edges like door posts, table legs, and so forth. The landmark precision is in most cases within centimeters. The TBF algorithm is implemented as a voting scheme, which group sonar measurements that are likely to have hit a mutual object in the environment. The algorithm has low complexity and is sufficiently fast for most mobile robot appli- cations. As a case study, we apply the TBF algorithm to robot pose tracking. The pose tracker is implemented as a classic extended Kalman filter, which use odometry readings for the prediction step and TBF data for measurement updates. The TBF data is matched to pre-recorded reference maps of landmarks in order to measure the robot pose. In corridors, complementary TBF data measure- ments from the walls are used to improve the orientation and po- sition estimate. Experiments demonstrate that the pose tracker is robust enough for handling kilometer distances in a large scale in- door environment containing a sufficiently dense landmark set. which is a prerequisite if triangulating objects with a small base line between the sensors. In this kind of research, high confi- dence classification of points, edges, and planes is reported (10), (12). The accuracy in position of the classified objects is claimed to be within millimeters, where the error comes down to being dependent on environmental factors like temperature, humidity and wind fluctuations. An appealing method, which is presented in this paper, is to manage with less signal processing, i.e., less range accu- racy, and still be able to reliably detect discriminating features, like vertical edges. The method we propose is called triangula- tion-based fusion (TBF) of sonar data, and buffers sonar scans that are triangulated against each other using a simple and low complexity voting scheme. Since the sonar scans in the buffer are taken from different robot positions, the base line between the sensor readings being triangulated are only limited by the odometry drift. Hence, the range accuracy is not of major impor- tance in this approach. This paper contains a detailed description of the TBF algorithm as well as a case study of how it can be applied for robot pose tracking. The paper is organized as fol- lows. Section II contains a detailed description of the TBF algo- rithm and discusses implementation issues. It is also explained how the TBF data uncertainty can be obtained using local grid maps. A real world example, where a sonar equipped mobile robot operates in a living room, is used throughout the section to illustrate the characteristics of the TBF algorithm. Section III describes how the TBF algorithm can be applied to robot pose tracking when using an extended Kalman filter. For related work, see (3), (12), and (16). A lot of details, such as tuned parameter values, are reported in order to document the implementation of the pose tracker. The idea is to use TBF data that have been validated against reference maps of TBF data for subsequent measurement of the robot pose. In corridors, the method is complemented with TBF data from the walls. A large scale indoor environment is used in the experiment to demon- strate the strength and weaknesses of the approach.

This publication has 18 references indexed in Scilit: