Name
Abstract | ||
|---|---|---|
Autonomous navigation in natural environment re- quires three-dimensional (3-D) scene representation and inter- pretation. High density laser-based sensing is commonly used to capture the geometry of the scene, producing large amount of 3-D points with variable spatial density. We proposed a terrain classification method using such data. The approach relies on the computation of local features in 3-D using a support volume and belongs, as such, to a larger class of computational problems where range searches are necessary. This operation on traditional data structure is very expensive and, in this paper, we present an approach to address this issue. The method relies on reusing already computed data as the terrain classification process progresses over the environment representation. We present results that show significant speed improvement using ladar data collected in various environments with a ground mobile robot. Recent advances in sensor design have enabled the use of laser radars that provide tens of thousands of 3-D points per second, even a hundred thousand, with centimeter range resolution. The problem now is how to handle such a large point cloud and how to design the data flow from the sensor to the environment interpretation. One critical aspect is t o be able to perform quickly basic operations such as insertion, access, and range search. Traditional optimal tree-based data structures are ill-suited for dynamic data sets because of numerous insertions produced by a moving robot in an outdoor environment. In this paper, we present a new approach for handling 3- D data for efficient on-board processing. The data processin g we are concerned with are kernel-based methods, where for a given point, some operations are performed using a support volume. The core of the approach is to minimize computation by re-using pre-computed intermediate results. The approach is demonstrated with data from a ground mobile robot, the Demo III XUV, for ladar-based terrain classification (14). The rest of the paper is divided into four sections where we present: the state of the art in data structures, specificall y for robot navigation; our approach, with a complexity and memory analysis; results from static ground robot, and the conclus ion. II. STATE OF THE ART |
Year | Venue | Keywords |
|---|---|---|
2005 | Robotics: Science and Systems | laser radar,data structure,data flow,dynamic data,point cloud,data collection,mobile robot,natural environment,three dimensional |
DocType | Citations | PageRank |
Conference | 4 | 1.13 |
References | Authors | |
6 | 3 | |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Jean-françois Lalonde | 1 | 590 | 37.69 |
| Nicolas Vandapel | 2 | 477 | 35.19 |
| Martial Hebert | 3 | 11277 | 1146.89 |