SBIR-STTR Award

We propose to study the feasibility of knowledge/geometry based mobile autonomous robot simulation system based on solid (3d Geometry) modeler for obstacles, a combined visibility free-space for robot knowledge, goal-finding exploration algorithms, and Facilities to specify these algorithms in terms of those modeling and representation schemes. A solid modeler facilitates complete representation of robot environment; visibility/free-space graphs corresponds to what the robot 'knows'. The separation of a priori known environment from the robot's accumulating knowledge entails Independent development and usage of object models and exploration Algorithms. Objects based on proven solid geometric modeling common in mechanical cad systems, can represent most obstacles of desired Shapes and complexity. A common type of solid model uses such Primitive solids as blocks, cylinder, cones, and spheres and form composite objects by performing Boolean operations (intersection, union, and difference) on them. Another common representation Scheme is based on boundary representation of all object surfaces. Objects with linear and quadric surfaces are easily represented. These thus encompass all polyhedra and polygons in a two-dimensional world. Visibility of obstacle features and the free space between obstacles represent the robot's knows through use of its sensors and suitable exploration processes. Various navigation algorithms, procedural or non-procedural AI techniques, can utilize these spatial knowledge effectively. A goal-seeking exploration algorithm will be further expanded to ascertain that it can be applied to the robot environment based on solid models.