[OpenRelief Developer] Modules and Drones - a hardware requirements note (was: Re: Modules - a hardware overview post)

[Shane Coughlan]

Hi Andrew 

On Mar 19, 2012, at 8:43 AM, Andrew Back wrote:
> I guess it would be good to have a better understanding of the
> end-to-end architecture including types of and number of nodes,
> distances, terrain, b/w required and power constraints etc.

The problem is that relief requirements will differ dramatically depending on the disaster. We can't be sure about the landscape or the number of modules in play. But we can develop designs for modules and transportation systems that allow a lot of flexibility.

== Situation example ==

Taking Tohoku as an illustrative example of a large tsunami, we can expect between 1 and 10km incursions of water that destroy virtually everything in their path for at least the first kilometer, and up to 500 kilometers of coastline impacted. All roads, communication systems and power supplies will be interrupted. An unknown number of people will be missing and dead, with the potential for hundreds to be lost in the water. Rescue efforts will be hindered by the destruction of infrastructure and the scale of damage. For example, the coastal towns and cities will suffer from massive churn, and wreckage potentially containing people will extend out to sea in a moving mass for several kilometers. This will all happen within the first couple of hours.

Initial rescue effort will focus on getting through to civilian concentrations, with towns and then villages targeted by aircraft. Emergency shelters will be contacted by land and air as quickly as possible to determine how many are intact, how many people are located in each one. and what their requirements are (in Tohoku about 110 of these shelters were overwhelmed by the tsunami with large loss of life). The initial outreach will take about three or more days with best effort, and only after that can secondary outreach to find people outside of these centers begin. 

Zooming in on the Ishinomaki peninsula as a 20km or so stretch of coastline, it would have been good to be able to deploy drones from Sendai (then Ishinomaki, then Onagawa as the army pushed through) to get a rapid picture of the situation, with the drones either broadcasting data back to base or recording the data for access on return. Automated visual identification of remaining roads, people and fires would have been good, but even standard video and photography would have been a big step forward, combined with data recording from one or more additional modules in the system (e.g. radiation). It would have been good to be able to drop powered modules over each town and village to monitor (for example) weather and radiation in an on-going manner. If each module could have sent a signal 500 meters, it could be reached by the drones on each overfly, and we'd have a functional emergency picture of that location. In that case we would have benefited from at least 2 modules (radiation and weather) per village and town, for a total of about 40 to 60 modules, with 10 or more drones to get the pictures and get the readings. 

 == What sounds sensible ==

I propose that two things make sense:
- The drones go as far as possible and be able to take different modules with them, as well as pick up signals from ground modules (our testbed drone will hopefully go about 20 to 30km in a 1.6 meter airframe).
- Modules are able work from the drones or work off other power-sources from the ground (AA batteries, car batteries). If using ground power-sources, they should transmit readings at least 200 meters and ideally 500 meters or so into into the sky and keep functioning as long as possible (low bandwidth but long life and reasonable transmission distance).

If these goals can be accomplished, then OpenRelief will have provided a way to accomplish a big step forward in disaster awareness. 

 == Summary ==

If OpenRelief compatible drones and modules can be launched then relief workers (a) cut down on the time to locate people and (b) get a better understanding of ground conditions and potential hazards (e.g radiation). However, the number of required modules/drones and other constraints is hard to predict. As an example location, Tohoku was full of small hills and terrible weather, plus the ground situation looked completely different to any previous mapping (just part of the problem was that a lot of land dropped 80cm, a substantial chunk moved almost 5 meters to one side). There is no way to fill the entire sky with drones or the entire land with monitoring tools to provide perfect clarity, but if we can overfly drones alongside standard ground and air services, and drop as many monitoring systems as possible that broadcast as long as possible, we will reduce the time to find and assist people.

Regards

Shane