Constrained Image Understanding for an Internet Robot Supporting Telepresence

The Network-Centric Applied Research Team (N-CART) introduces a distributed model for constrained image recognition for a web-enabled video robot supporting aspects of telepresence. The recognition engine resides on a network node apart from the actual robot streaming the video. We have found that by employing such a model, the cost and complexity of the robot can be reduced while allowing the recognition services to be reused by other devices. At the end of the paper, we discuss how we are using the notion of distributed services to deploy video preprocessing for limited capability wireless devices.