Home

3. Multi-layered Based Interaction and Navigation for an Intelligent wheelchair [paper].

In a semantic map, different places and objects are classified and categorised, such as common room, living room, kitchen, desk, bed, furniture, food etc. Each place and object has their own properties namely class, name, ID, function, pose, state. It also represents the topological relation of the components in the environment. Semantic spatial representation can enable a robot to interact with human through natural language and corresponding linguistics based label or text interface contains this information. A robot with semantic representation of heterogeneous places and objects can augment human-robot communication and interaction. Additionally, the semantic information can improve the efficiency of path planning and navigation as well as extend the capability of task planning for a robot. This paper presents an interactive navigation system for an intelligent wheelchair with a multi-layered map prototype mentioned above. The geometric map provides the spatial information for localisation. The topological map and geometric map are used for path planning. The topological map consist of segments of environment graphs. The semantic map representing the environment with an ontology model is used for human-robot interaction and task planning. Dialogue is used to provide a natural and friendly interface for human-wheelchair communication.