Pervasive Surveillance Networks

1
Pervasive Surveillance Networks
Clayton G. Haffner, Chad Klochko, Michael
Huntwork, Craig Pomeroy, Greg Moy Professor
Mutka, Professor Xi, Amit Goradia

• Tele-operation
• Physical movement of sensor based on commands
  issued by operator, by mouse or joystick.
• Allows for hands-on approach to surveillance, and
  gives operator an active role in the surveillance
  task.

• The Surveillance Task
• Maintain constant sensor coverage of a given
  area, and inform operators of changes within the
  area.
• Discerning Targets, Identifying what from the
  viewable region is a potential target.
• Maintaining Targets
• Current Surveillance Techniques
• Multiple Homogeneous Sensors with little or no
  range of motion.
• Multiple monitors, with few operators monitoring
  them.
• Problems with Current Approach
• Require a large number of sensors.
• System is static, not dynamic
• Operator is a passive observer to the system, has
  to active input.

• Experimentation Results
• Set-up
• Multiple Heterogeneous Sensors.
• A heavily modified version of the Palantir
  client, for a video server.
• Experiment 1 Tele-operation
• Operator has complete control over camera through
  mouse commands.
• Experiment 2 Pervasive System
• System chooses a target then maintains that
  target based on tracking criterion.
• Robot follows target to maintain adequate
  resolution.

• Mobile Sensors
• Sensors with either full motion capability or,
  the ability to change their viewing region.

• System Decision Making, solving the Surveillance
  Task
• System chooses targets based on color
  segmentation scheme, implemented using CMVision.
• System uses Hausdorff Tracking to maintain the
  target not only within visibility but also within
  the best possible resolution and coverage.
• Hausdorff Tracking
• Treat target as a set of points, then attempt to
  minimize the distance between the coverage set,
  and the target set.

• Goals for the Future
• New Camera Client
• Implementation with different Heterogeneous
  Sensors.
• Experimenting on a larger scale to take advantage
  of the full range of motion of the sensors.

Information provided by M. Huntwork, A. Goradia,
N. Xi, C. Haffner, C. Klochko and M.
Mutka,Pervasive Surveillance Using a
Cooperative Mobile Sensor Network,''Proceedings
of IEEE International Conference on Robotics and
Automation(ICRA 2006), May 2006. C. Haffner, C.
Pomeroy, M.Huntwork, G.Moy
. Modeling Design of
Pervasive Surveillance Networks, 3rd Place
Winner of International Hands-on Competition at
National Chung Cheng University in Taiwan.