UBC LCI: Vision as a Human-Robot Interaction modality - Underwater Robotic Apps

You're invited to attend a talk by a PhD candidate from McGill University, Junaed Sattar, who will be visiting UBC early next week. Junaed will give his talk during the LCI Forum slot (even though the forum is canceled for summer), for about half an hour plus questions. Please invite non-LCI members who might be interested as well.

Abstract:
This talk looks at a vision-based scheme for human robot interaction in an underwater environment. Interactions are classified as explicit and implicit interactions. In the explicit mode, the operator (often a scuba diver accompanying a robot) instructs the robot with gestures formed using a pair of engineered markers, or fiducials. These gestures are interpreted as commands, which can be both atomic and compound in nature. We define a visual language called RoboChat, that enforces a strict grammar on the types of sentences that can be formed. In the first half of this talk, we introduce our amphibious robot, Aqua, and briefly look into the structure of RoboChat and also look into some experimental data from field trials.

In the second half, we present an algorithm for underwater robots to visually detect and track human motion. Our objective is to enable
human-robot interaction by allowing a robot to follow behind a human moving in (up to) six degrees of freedom. In particular, we
have developed a system to allow a robot to detect, track and follow a scuba diver by using frequency-domain detection of
biological motion patterns. This falls into the general category of implicit interactions. The motion of biological entities is characterized by combinations of periodic motions which are inherently distinctive. This is especially true of human swimmers. By using the frequency-space response of spatial signals over a number of video frames, we attempt to identify signatures pertaining to biological motion. This technique is applied to track
scuba divers in underwater domains, typically with the robot swimming behind the diver. The algorithm is able to detect a
range of motions, which include motion directly away from or towards the camera. The motion of the diver relative to the vehicle is
then tracked using an Unscented Kalman Filter (UKF), an approach for non-linear estimation. The efficiency of our approach makes it
attractive for real-time application on-board our underwater vehicle, and in future applications we intend to track scuba divers in
real-time with the robot. The paper presents an algorithmic overview of our approach, together with experimental evaluation based
on underwater video footage.

No more details on this talk.

For details on the LCI forums in general see:
http://www.cs.ubc.ca/...