Questions and information about using technology to support wilderness search and rescue.

Ron Zeeman is an incident commander-level volunteer with Utah County Search and Rescue. He has been the subject matter expert for all of the NSF-sponsored work on using UAVs to support wilderness search and rescue, including leading most of the field trials. He is also developing technology to support underwater search, unmanned helicopter-based search, and computer vision from manned aircraft.

• Can we give a demo at MRT? MRT is a large training exercise, and it might be good to demonstration how UAVs can be used to support search and rescue. It would also be a great opportunity to demonstrate other technologies, and get some feedback from other subject matter experts.
• What would be required to use the Wonder Server computer vision toolbox it manned aerial search, that is, to record and analyze video obtained from a manned UAV?

Craig Snyder is from the Greater Philadelphia Search and Rescue group located outside of Philly. We are a SAR team serving SE Pa, and parts of NJ and DE. We have been around since 1976 and provide trained ground searchers, K-9 teams, and search managers for searches headed up primarily by law enforcement and fire departments. They are currently working on providing aerial capabilities to the team. They work in a mixed wilderness/suburban/urban area. Our needs are slightly different then what the western teams are probably looking for. Many of their searches are also probably of a much different nature. They are looking for aerial support for initial recon of the area, identification of hazards, unmapped structures, terrain type, and high probability targets. They also want to be able to monitor the progress of their tracking teams and K-9 task teams and to monitor the area immediately in front of them. They often deal with people who don't necessarily want to be found and they actively avoid the search teams; note that these are not criminals, but rather people who are mentally impaired or scared for one reason or another..

• Can we provide a parts list? I don't see why we can't provide what we've done in the past and share with him the things we are planning on doing in the near future.
• What video transmitter are we using and what is its range?
• What are reasonable flight times to expect from
  • A hobby-style fixed wing aircraft
  • A WiSAR-specific airframe
  • A fixed-wing aircraft from Procerus
  • Hobby-style quadrotors
  • Gas-powered helicopter UAVs
  • Gas-powered fixed-wing UAVs
• (Actually, this is a derivative of some points that he made.) If we create a software tool to help support SAR, it really needs to reinforce good ICS practices.
  • How do we make this possible?
  • It probably isn't necessary to include predicting lost person movement for many kinds of searches.
• What are the best practices for small search areas, that is, for areas where the missing person is likely to be found hurt very near their expected routes of travel?
• Are their camera technologies that use special images/wavelengths technologies to penetrate a forest canopy?
  • If not, what could be done with a quadrotor to get good angles to obliquely peer under the canopy?
  • Could you use a quadrotor to fly along a trail under the canopy?
• What technologies would make it possible for a UAV to help in water searches? Note that they are looking for submerged victims.
  • Are there wavelengths that can see through water and silt?
  • Are there wavelengths that could sense heat differentials?
  • What kind of computer vision technologies could be used detect anomalies or patterns?
  • What kind of affordable, autonomous underwater vehicles could be used?
• (Another derivative question …) What can we do to use UAVS to help with pre-planning, that is, initial recon, identification of possible hazards, and places to put ground searchers?
• (Ditto …) What can we do to help K-9 and other teams monitor the area immediately in front of them?
• Craig makes the following great suggestion. “We currently have our team, and one our members who happens to work at a junkyard, keeping their eye out for smashed up BMW's and Mercedes. They include a heads up display with a night time IR camera that is supposed to be able to “see” 200-300 ft. ahead. This would be more than enough distance for a drone flying at roughly 75-150 ft. They have aftermarket kits that retail around 1500. We can't afford that and all we want is the camera itself any way. We don't want all the wiring harness, monitor, etc that comes with it. We'll kit bash it and figure out how to power it and get the video out. We are looking into trying to get a dealer to donate a kit at cost or help us get one from an accident. Unfortunately since the accidents that usually total cars are head on, the cameras are usually toast as well. Anyway, thought your group might also be able to get a local dealer to donate an IR kit to your group to use in the research. We like doing night searches, especially with the air scent dog teams since most lost people remain stationary at night and scent can build in an area. We also find most civilian volunteers, police and fire personnel aren't keen on night searches and often aren't equipped for them, which makes it a lot easier for us to search with the dogs since people aren't out there.”

Walter Arlt is a retiree who works with his son on a variety of technologies for supporting human welfare. Their company has recently been working on developing UAVs for use in various applications including search and rescue. Videos of their work can be found here.

Nelson Trichler is from Santa Barbara County Search & Rescue. Their overall operations naturally rely on their experience, both in what subjects have done in the past and their knowledge in the areas someone is lost. The latter requires getting out and seeing what the terrain is like: Is it flat ground with little vegetation or mountains with think chaparral that limits one's ability to move through therefore they tend to stay on established trails? Visibility is a key factor in their searches. If someone is in the wilderness and cannot see any specific reference points then they might go anywhere. On the west coast many of the coastal mountains overlook the ocean so most persons know that if you go downhill you will eventually run into the coast and other people. The trouble is the vegetation and the ability to get through it if the person is off trail.

• How do you incorporate those factors like terrain and vegetation into the input process?
• What mapping software is being used?
  • How does this compare to Terrain Navigator Pro from MyTopo which is relatively inexpensive but has a lot of tools SAR personnel can apply such as real time tracking of field teams, 3-D mapping, ability to document POAs and PODs, etc.

Don Ferguson is a well-respected volunteer who has been involved with search and rescue for over 10 years. He has contributed significantly to an understanding that efficient SAR planning involves a lot more than just have a lot of people walking around in the woods. Understanding how individuals interact with their environment plays a key role in SAR planning, and is very familiar with Robert Koester's work.

• How are we using or planning to use Robert Koester's Lost Person Behavior models including his discussion of Bayesian models?
• How are we using or planning to use the International Search and Rescue Incident Database (ISRID)?
• What could we do to contribute a behavior element to the model that he has been working on in conjunction with ESRI, the National Park Service, and members of the SAR community?

Steve Campbell is working to bring together a project team using ArcGIS to build on our significant lost person behaviour data. The ESRI team here is talking with staff at Yosemite National Park who are doing some work too and we are hoping to build on their computer models.

• Is there a way that we could contribute to his work?

Charles Twardy is a researcher at George Mason university with expertise in search theory and analysis of lost person behavior. We are currently working with him to try and obtain funding that will allow him to lead an effort to evaluate the quality of various searcn maps and models. Charles and Bryan Morse have made some funding available that may allow us to get some of Robert Koester's time to consult on this project.

Robert Koester is both an Incident Commander and a researcher interested in Lost Person Behavior, Sweep Width, and general SAR theory. He has done extensive work on using past data to help identify probability of area (POA) models to guide search. Lanny's work might be able to add something to Bob's work by looking at how a particular individual is likely to travel through a specific terrain of interest. Ideally, it would be best to find a way to integrate Lanny's work into Bob's excellent foundation work.