Multi-modal Person Tracking: Toward Implicit Communication for Peer-to-Peer Human-Robot Teams in Shared Workspaces
Scott C. Livingston · 2009
I present here a multi-modal system for person tracking designed for mobile robot platforms equipped with laser range measurement and color camera sensors. Basic person tracking is achieved by filtering “leg” circle features extracted from a laser sensor, considered in combination with the target position (with respect to a particular image) estimated by a vision-based tracking algorithm. This work implements and builds upon previous research successes, and is novel primarily in its combination of the individual elements. The final project strives to be easy to use, being documented and based upon an object-oriented design, and should thus represent a nontrivial contribution toward a larger research project to develop methods for peer-to-peer human-robot teams through implicit communication.