Just-In-Time Sensing

Mobile Robot Exploration
Using Fused Data from Two Sensors.

Abstract


This research project investigates the combined use of a sonar range finder and a laser range finder (Biris) for exploring a structured indoor environment (i.e. an art gallery). The environment of an art gallery is particularly attractive for implementing a mapping since it is a generally open space. It has a number of stationary objects and allows full access to the space around them for the visitors, as well as for a mobile robot of a size comparable to that of a human. The assumption is that the floors are flat (except for some well defined places where stairs exists) and smooth without any anomalies. Moreover the area is free from moving objects (such as humans) for long periods of time.

This approach fuses the data acquired from two different kinds of sensors in order to compensate for the unavoidable disadvantages any single kind of sensor presents. The first sensor is a sonar range finder that is coarse, but covers a longer range and it is relatively fast, and the second much slower is a laser range finder that can be used to report precise information about a small area. This combination of sensors is going to be used by taking first a coarse estimation of the environment, and then ``cleaning up'' the uncertainties with the laser sensor.

The main disadvantage of a sonar range finder is the possible false estimate of distance caused from multiple bounces. Also, the way distance is estimated by it leads to uncertainty at the boundaries of objects. In order to enhance the performance of the sonar range finder and to construct an accurate map, we combine it with the Biris laser range finder, taking measures at the ends of the line segments to remove any uncertainty. Also erroneous data from multiple bounces are going to be eliminated by close examination with Biris.

The space exploration is based on two strategies. In the first, a rather coarse map, clear enough (to allow secure navigation through the free space) is constructed using the sonar range finder, and then the Biris sensor is used in order to investigate uncertain areas that present a threat of collision. In the second approach the objective is to accurately map the environment, and the precision of the construction is going to take precedence over the time spent in the task. In this case the sonar is used to identify areas where could be an object, and then the robot approaches and creates a detailed map using Biris.

References

  1. Environment exploration using ``just-in-time'' sensor fusion

    Gregory Dudek, Paul Freedman and Ioannis M. Rekleitis
    Vision Interface, pages 175-182, Toronto, May 1996

  2. Just-in-time sensing: efficiently combining sonar and laser range data for exploring unknown worlds

    Gregory Dudek, Paul Freedman and Ioannis M. Rekleitis
    Proc. of IEEE International Conference in Robotics and Automation, volume 1, pages 667-671, Apr. 1996


Complete List of Publications and bibtex entries.




Back to my CIM HOME PAGE