SPIE Startup Challenge 2015 Founding Partner - JENOPTIK Get updates from SPIE Newsroom
  • Newsroom Home
  • Astronomy
  • Biomedical Optics & Medical Imaging
  • Defense & Security
  • Electronic Imaging & Signal Processing
  • Illumination & Displays
  • Lasers & Sources
  • Micro/Nano Lithography
  • Nanotechnology
  • Optical Design & Engineering
  • Optoelectronics & Communications
  • Remote Sensing
  • Sensing & Measurement
  • Solar & Alternative Energy
  • Sign up for Newsroom E-Alerts
  • Information for:

SPIE Photonics West 2017 | Register Today

SPIE Defense + Commercial Sensing 2017 | Call for Papers

Get Down (loaded) - SPIE Journals OPEN ACCESS


Print PageEmail PageView PDF

Defense & Security

Multi-threat containment with autonomous formation of robot teams

Robots monitoring hazards cooperatively form dynamic groups to surround and contain threats.
11 July 2008, SPIE Newsroom. DOI: 10.1117/2.1200807.1194

Technological advances in many scientific fields have enabled cost-effective robots to perform tasks cooperatively and without human supervision. For example, distributed swarmlike behaviors and wireless communication provide opportunities to solve problems such as toxic chemical spills and wildfire containment, hazardous scenarios where human intervention would be risky or even impossible. Coupling wireless networking and simple yet efficient swarm intelligence, a truly decentralized solution has now been developed where robots dynamically form cooperative teams to tackle multiple threats or minimize their impact quickly and efficiently. Figure 1 shows an example of three threats contained by teams of robots.

Robot teams have been tried in terrain exploration,1 target tracking,2 target pursuit3 and the coordinated movement of objects.4,5 More recently, the containment of immobile and mobile threats has drawn attention to the potential of cooperative robotics. For example, Cui and colleagues used robot swarms to localize the emission source of an aerosol threat, but required the robots to work as a single team and to treat only a single source. The group of Stancil considered multiple threats, but the robot actions were coordinated by a central controller processing images transmitted from cameras mounted on each robot.

Facing multiple threats in a potentially widespread terrain, each robot in our solution uses its sensors to gather local information, exchange wireless messages, make autonomous decisions to establish or join a threat containment team, and then dissolve the team after the threat is successfully contained.

Figure 1. A multi-threat containment example. A subset of robots (•) surround three threats (*).
Enablers for distributed multi-threat containment

Key elements of an autonomous, efficient, and robust threat-containment strategy include use of the so-called random direction mobility pattern before any threats are detected. This has been shown to avoid a wasteful concentration of entities (robots) in central areas and provides a more uniform coverage of the search space.6 When searching for threats, limited robot sensor ranges can also be extended by using a wireless broadcasting protocol so robots follow the angle of any received ‘call-for-help’ (CFH) signal. If more than one CFH signal is received, a robot will compare them to threats detected by its own sensors and move toward the closest target.

When a threat is targeted, collaborative movement of robots to surround the threat is crucial. To accomplish this goal, each robot finds the mid-angle between neighbors to its immediate right and left and moves toward that angle while maintaining a safe distance from the threat by creating an ‘artificial potential field.’ A robot will consider the threat to have been contained provided a suitable distance and angle are maintained between itself and its neighboring machines (see Figure 2).

Finally, when more than one threat is presented, a local robot neighborhood is established when a time division multiple access (TDMA) channel is dynamically created by the robots for a particular threat. The autonomously elected leader of the TDMA neighborhood will transmit the CFH signal as well as command the team to disband if the threat is perceived to be contained by all members.

Figure 2. Each robot will move toward the mid-angle of its immediate neighbors while maintaining a proper distance from the threat using artificial potential fields.

Figure 3. Robot prototypes.

The integrated solution7,8 based on the elements described above allows effective coordination of robots toward the same threat, even if the threat is outside an individual robot's local sensing range.

Prototype robots and results

Four prototype robots are shown in Figure 3. A modular event-driven simulator has also been developed to evaluate the performance of the proposed solution. To our knowledge there are no other reported results on multiple threat containment. Table 1 shows the simulation parameters and the average performance achieved with our system over 10 independent runs each lasting 6000s.

The results show that 92–95% of threats are successfully contained within 18s. This impressive performance is sustained as long as there are sufficient robots in the search space and they are adequately equipped. The robots' performance parameters allow small, cost-effective machines, and our experience developing the prototypes suggests a minimum per-unit cost of approximately $500. A low unit cost per robot and a distributed solution make it possible to deploy large numbers of units and have the overall system sustain a good performance, even if a few robots fail.

Future research

The solution described is the first of its kind to tackle the dynamic, multi-threat containment problem. Our emphasis provides a truly distributed system where autonomous robots dynamically form teams to contain multiple threats. Our simulation results, even using moderate robot performance parameters, demonstrate a promising 92–95% containment rate with the average time to contain value in the range of tens of seconds. This research is now being extended to treat scenarios where threats may react to containment by escaping or disabling the robots.

Table 1. Simulation parameters and results.

Shanchieh Jay Yang
Department of Computer Engineering
Rochester Institute of Technology (RIT)
Rochester, NY

Shanchieh Jay Yang received his PhD from the University of Texas, Austin, in 2001 and is currently with the Department of Computer Engineering at RIT. His research interests include cyber threats, impact assessments, and cooperative robot-sensor networks. He was a co-chair for the IEEE Joint Communications and Aerospace chapter in Rochester, NY.

Nathan Ransom
Harris Corporation
Rochester, NY

Nathan A. Ransom received his BS and MS degrees from Rochester Institute of Technology in 2004 and 2008, respectively. His research interests include robotics, swarm theory, and wireless networks. He is currently employed by Harris Corporation's RF Communications Division as an embedded software engineer.

Bhushan Mehendale
Microsoft Corporation
Redmond, WA

Bhushan U. Mehendale received his BSc/MSc dual degree with highest honors in computer engineering from RIT in 2006. His research interests include robotics, image processing, security algorithms, control systems, and computer simulations. He is currently a developer at Microsoft Corp., Redmond, WA, in the Windows Mobile®Core OS group.