We talked about five specific areas of research and distilled them into a theme for 2011: Multi-Flight, Extended Search WiSAR. The idea is that our work to date has focused on pretty short-term searches – the kinds of flight paths that you might execute and the kinds of data you might consider in one, 75 minute flight. In reality, it's very unlikely that a search will be complete in 75 minutes because we probably won't find the missing person in the first flight. So, the big research question for the year is what do we need to do if the search goes on for a couple of days or requires multiple flights?

• I suspect that we've probably learned about all that we can learn from the simple field tests that we've been doing. How many more times do we need to fly a lawnmower pattern over a likely search area until we see something in the video.
• But I also believe that there is much to be learned from the field. We need to improve the field worthiness of the technology before it could be practical, and we also have much to learn from a scientific perspective.
• We should shift from “field test” to “mission test” mentality, with emphasis on planning and integrating multiple flights and ground search efforts.

• Incident commander interface
  • Use Mike R.'s 2D Google Earth replacement as base for Incident commander interface.
  • Missing persons modeling (Lanny)
  • GPS quality estimation (Spencer)
  • Seeability from ground searches (Mike R.)
    • I'd like to see my PC code ported to Android devices so it can be used during searches. Integration of Spencer's GPS quality estimation would also be beneficial.
  • It seems like the UAV piloting interface and the IC interface should maybe be the same. -Spencer
• Exploiting shadow information in assisting search and rescue (Josh A.)
  • Finding the lost persons through shadow classification
  • Shadow segmentation and removal
  • Exploiting shadows to correct height maps
• Mixed resolution imagery over multiple temporal frames
  • Limited comms channel – which information to send and what should be done with the UAV while sending it? (?)
  • Temporal frames include online search, offline search (e.g., after the day's searches are done), and between flights search. (?)
  • Temporal integration of information from, for example, shadows.
• Outreach
  • Charles Twardy and R. Koester will be working on incident commander map ideas.
  • Don Ferguson has expressed interest in including the missing persons model in his IC interfaces.
  • Santa Barbara and Pittsburgh (?) have expressed interest in learning what we've done and in using some of our software.
  • Zoe Wood (Cal Poly) is interested in predicting how humans move across the terrain in archaeological applications.
  • Mike G. needs to work with Ron and follow up on getting FAA approval so that Utah County can do real searches with the technology.
  • We need a path to produce sustainable technology for real search.
    • Most SAR groups cannot afford anything remotely as expensive as the equipment we get from Procerus. It's more likely that they'd use something cheaper. They're going to fly them into cliffs in the rain. DIYDrones is a community of hobbyists that build their own UAVs. They've got a cheap (a couple hundred dollars) arduino-based autopilot solution. I (Spencer) envision SAR teams flying cheap RC planes outfitted with this kind of autopilot. Total cost for a plane with autopilot and camera could almost certainly be <$2000. The downside is that it would not be an integrated solution like the Procerus equipment.
    • Our software and methods should be made platform independent (within reason).

• Wonderbox version 2 should integrate coverage maps and infrared technologies, and support a system where two people can watch videos at different timeframes using different tools.
• Lanny's path planning algorithms need to be flown.
• Lanny's areas of focus map and task-difficulty map modification tool using touch-screen gestures need to be developed and field tested.
• Phairwell needs significant refactoring or needs to be thrown away altogether. (Possibly integrate that functionality with upcoming IC interface)?