A framework for visual landmark identification based on projective and point-permutation invariant vectors
✍ Scribed by Christos I. Colios; Panos E. Trahanias
- Publisher
- Elsevier Science
- Year
- 2001
- Tongue
- English
- Weight
- 602 KB
- Volume
- 35
- Category
- Article
- ISSN
- 0921-8890
No coin nor oath required. For personal study only.
✦ Synopsis
Autonomous robotic navigation based on machine vision, presents a challenging field in current robotics research. This task presupposes a representation of the environment and methods to process this representation and use it to derive the appropriate platform motions, in order to accomplish specific navigation goals. Following recent theories of active and purposive vision, we attempt to qualitatively describe the environment, avoiding thus a metric, 3D representation of it. A set of reference patterns, the so-called landmarks, together with their topological relations, is used to adequately describe the robot's workspace. Mathematical tools from projective geometry are employed for landmark identification, facilitating robust landmark recognition irrespective from the camera viewpoint. A complete framework is presented in this work for landmark extraction and recognition based on projective and point-permutation invariant vectors. Detailed experimentation has revealed accurate landmark recognition in indoor workspaces.