Sonar-based real world mapping system.

Abstract

Autonomous robot navigation is a complex processing task which can be broken down in several levels. These levels are Robot Control, Sensor Interpretation, Sensor Integration, Real-World Modelling, Navigation, Global Planning and Control. The work done in this thesis addresses the first four of these levels in order to generate an accurate environment model from sensory data. A real world mapping system was developed and implemented for a mobile platform evolving in an unstructured environment. The system uses sonars as the source of sensory data. A certainty-grid based mapping method is used to map and merge the sonar data onto a pixel representation. For the purpose of this research, a user-friendly application is used to provide a test platform and demonstrate the system capabilities.