Session: 07-11-03 Mobile Robots and Unmanned Ground Vehicles III

Paper Number: 70452

Start Time: Tuesday, 07:00 PM

70452 - On the Mapping Problem in SLAM Approaches for Autonomous Robot Navigation 

Simultaneous Localization and Mapping (SLAM ) is a well know strategy for enabling robots to maneuver in unknown environments.  The solution to a SLAM problem provides robots with information of where they are, and the structure of the environment in a way that enables the robot to make better decisions on their next move. As indicated  in the name, SLAM performs two tasks at a time: determining the location of the robot, and describing the map of the environment. There are many ways of localizing the robot by using on board sensors, which can include ranging sensors, motion sensors and vision sensors; however all of them tend to share the mapping strategy. Typically, there are two  main methods of creating the map. One approach augments the coordinates of all landmarks in the environment with the robot pose as one state vector, and the second creates a 2-D or 3-D grid of the environment and then assigns each grid location the occupancy level of the landmarks. Although both methods have been very effective, they do have some limitations. The first method tends to suffer a problem of increasing the size of the state vector continuously, which can be computationally demanding, and the second has a problem treating all grid occupancy levels in binary form where all occupied grids are assumed to be at equal levels while it is possible that certain grids my have occupancies that have no effect on the mobility of the robot. This paper reviews the mapping methods used by most SLAM algorithms and presents a simple but effective grid based mapping strategy that integrates the path planning step for autonomous ground robotic vehicles. The map is presented as a weighted 2-D grid in which each grid is continuously weighted. The weights are determined based on the observed locations of landmarks, their sizes, and distances from those landmarks, where grids close to the landmarks have more weights than those that are far away from the landmark. These weights are updated continuously without affecting the size of the map in memory although some grids will have more weights than others. The path planning task is implemented by joining contiguous grids that have less weight. Preliminary performance results will be presented by comparing this map and the one formed by binary occupancy level in terms of the path planning results. It will be shown that under the path binary occupancy map, path planning is more computationally intensive compared to the proposed map.

Presenting Author: Vomsheendhur Raju North Dakota State University

Authors:

Vomsheendhur Raju North Dakota State University
Majura F. Selekwa North Dakota State University