Sensor-based discovery path planning is problematic because the path needs to be continually re-computed as new information is discovered. A process based client-server approach has been successfully deployed to solve this problem, thus permitting concurrent sensor-based map and localization-correction updates as well as concurrent path computation and execution. A potential function is created by solving Laplace's equation using an iteration kernel convolution with an occupancy-grid representation of the current free space. The path produced is optimal, i.e., minimizing the distance to the goal in addition to minimizing the hitting probability. In this paper the approach is extended to include a specification on the target approach: appropriately referred to as the serving component of the cocktail party problem. The extension is unique in that no additional computational complexity is necessary. The path is only biased when open space is detected between the current robot position and the goal.
展开▼
