Name
Abstract | ||
|---|---|---|
This paper presents an approach to explore an unknown indoor environment using vision as the sensing modality, thereby building a topological map of images. The contribution of this paper is in a new approach that identifies the next best place to move from a node in the topological graph. This decision is taken locally at a node by choosing the next best direction, when there are open spaces before the robot, and globally by choosing the next best node to branch off a new exploration, when there are no open spaces before the robot. We propose a method to assign weights to nodes for this purpose. Weight is defined as a function of the depth of local descriptors of images, and the number of times they were seen across different nodes. The efficacy of the approach to explore office like environments is verified through several experiments on a P3DX robot. |
Year | DOI | Venue |
|---|---|---|
2010 | 10.5555/1838206.1838450 | AAMAS |
Keywords | Field | DocType |
topological graph,different node,new exploration,open space,next best direction,indoor environment,new approach,next best node,local feature,topological map,next best place,p3dx robot | Computer vision,Computer science,Image based,Artificial intelligence,Topological map,Topological mapping,Robot,Topological graph | Conference |
Citations | PageRank | References |
0 | 0.34 | 11 |
Authors | ||
3 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Aravindhan K. Krishnan | 1 | 1 | 1.41 |
| Madhava Krishna | 2 | 13 | 2.12 |
| Supreeth Achar | 3 | 132 | 9.10 |