[Osmf-talk] Notes & Bing Imagery

John Crowley bostoncello at yahoo.com
Thu Dec 5 15:20:30 UTC 2013


As someone who works in crisis response, I grok the need for fresh imagery. It is one of my biggest pain points. Like many, I look forward to getting imagery quickly from platforms that can collect wide areas quickly. I also want them to carry communications gear that restore internet and voice comms. It's why we run an experimentation program every quarter.

That said, there are a couple assumptions here which I have to correct. 

1. Autopilots and preprogrammed routes have large error bars. I have observed a comparable (more expensive) version of SenseFly wander 100s of meters off the planned route. In a forest survey, that is a problem (vertical obstructions do matter, but altitude provides some safety). In a city, where tall buildings and radio towers mix with general and commercial aviation, it is a major risk to veer that far off a planned route. We need to stop believing the marketing hype and start working with the platforms in places/sites where it is safe to fail. Cities are not on my list of such sites.

2. Small <> harmless. Just because a UAV has low mass does not mean it cannot cause harm. First, the weight class you describe <>3kg) is about the same as a bird. Take a peek at the US Federal Aviation Administration (FAA) reports on the effects of bird strikes, most of which happen under 150m above ground level with animals from 100g to 3kg (see http://wildlife.faa.gov/downloads/StrikeReport1990-2012.pdf for a detailed summary of the negative effects). Second, fixed wing UAVs generally don't fall from the sky; they crash at velocities exceeding 30km/hr. An 1kg object moving at this velocity with a spinning prop can hurt someone up pretty badly if they are struck. I've seen it happen at model airplane shows.

3. Sophisticated software <> fail safe UAV. Yes, robots can react faster than a pilot on the ground to changing conditions (the OODA loop is much shorter). But, the aircraft you describe have to compromise cost and capability. That ratio is changing, but from what I have seen, deviations from ideal conditions lead to unexpected results. It makes sense. Sensors and software at this price point are a matter of finding good compromises. These compromises do not cover every potential disruption--not just weather, but an obstacle that is either not yet mapped or which moves in front of the UAV. Given our use case (mapping after disasters), the map is by definition different from before the event and lots of helicopter traffic moves in sometimes ad hoc flight paths. It's not a great fit. I would believe less marketing documents and work more with the UAVs themselves. You'll see what I mean. There are lots of engineering compromises that marketing glosses over.

For the foreseeable future, I envision manned aircraft with UAV sensors mounted on the undercarriage will be the hybrid that allows for our work to move forward. We should experiment with UAVs, while building better relationships with general aviation pilots who can fly our sensors where the UAVs cannot (and should not) yet go. 






On Thursday, December 5, 2013 7:42 AM, Martin Koppenhoefer <dieterdreist at gmail.com> wrote:
 


2013/12/4 Simon Poole <simon at poole.ch>

Well reality is somewhere between your scenario and being hit by a one kilo blanket dropped from 100m altitude. It would surely hurt, so much is sure. 
>

Yes, I agree ;-)


Anyway, German regulations currently seem to allow drones with a mass of up to 5kg to be flewn without approvation.


Yes, there are also other risks like being hit by a falling tree or by a falling roof tile. For example a (classical German) roof tile has a mass of ~4.3 kg and falling from 20 m would usually kill you if it hits you on the head.


cheers,
Martin


_______________________________________________
osmf-talk mailing list
osmf-talk at openstreetmap.org
https://lists.openstreetmap.org/listinfo/osmf-talk
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstreetmap.org/pipermail/osmf-talk/attachments/20131205/e372b66e/attachment.html>


More information about the osmf-talk mailing list