4 ms·
This is great! Sounds like they are solving the problem of detecting tiny fires in vast wilderness[0]. Detection seems like 80% of the problem. Immediate respo
by tobyjsullivan 5y ago
This is great! Sounds like they are solving the problem of detecting tiny fires in vast wilderness[0].
Detection seems like 80% of the problem. Immediate response would be a huge improvement over the current strategy. That's a win even if they don't solve the drone scaling/deployment/automation problems.
[0] https://www.rain.aero/missions/software-engineer-realtime-fire-mapping https://www.rain.aero/missions/software-engineer-realtime-fi...
- svnt 5y agoIt looks like a practical use-case for drones even if the dropping of suppressive poke-balls probably needs improvement. Rough numbers: 3 million acres of forest in USA, one drone could ultimately cover 500 sq km (13 km radius) 6000ish drones call it 10k for redundancy Even at 10k per drone including base station that’s roughly $100M in hardware to cover the entirety of US forestland. PG&E is your first customer. edit: Even if I’m low by an order of magnitude on costs, PG&E is your first customer: https://www.utilitydive.com/news/cpuc-imposes-largest-ever-penalty-of-19b-on-pge-for-northern-california/577625/ https://www.utilitydive.com/news/cpuc-imposes-largest-ever-p...
- python999 5y agoMaybe they would just monitor the corridors along transmission lines. That would be a tiny fraction of the area.
- byw 5y agoHonestly $100M is not bad considering a single Boeing 737 costs the same amount. But realistically you'd want winged drones, otherwise you'd need a large number of bases (power hookups, maintenance, etc.).
- ted_dunning 5y agoYellowstone park alone is over 2 million acres. 3 million acres is a wild under-estimate. You are nearly TWO orders of magnitude off.
- realslimjd 5y agoThere are 750 million acres of forest in the US and each drone can cover 40 acres. That's 18,000,000 drones required.
- Jabbles 5y agoThey aren't actually detecting the fires, according to this article linked on their press page: Each drone would be parked inside a secure shelter and launched in response to fire-watch cameras or satellite imagery. It would navigate autonomously to the ignition site using “computer vision,” its infrared sensors feeding data to AI software to estimate the size, growth rate, and direction of the fire, before “deciding” whether to drop retardant on the fire, create a firebreak ahead of it, or summon more drones. Human operators would oversee and approve its operations. https://www.mercurynews.com/2020/11/30/can-bay-area-startups-firefighting-drones-save-us-from-catastrophe/ https://www.mercurynews.com/2020/11/30/can-bay-area-startups...
- quarterdime 5y agoIt seems to me like they are solving the problem of detecting tiny fires in sections of 40 acres each [1]. Yosemite is about 700k acres. [1] https://www.businessinsider.com/startups-working-to-fight-wildfires-climate-change-2020-10#rains-drones-are-like-sprinklers-for-the-forest-3 https://www.businessinsider.com/startups-working-to-fight-wi...
- cjensen 5y agoSatellites are already used to detect hotspots. The Alert camera network [1] catches a lot too. On days where fire would be dangerous, there are already planes ready to go to get the fire before it gets large. At best, this would provide more coverage more quickly by forward deploying drones. But planes are fast enough and you only need to use a smaller plane on a small fire, and we have plenty of them. Sure seems like techies had a bright idea who either failed to do the research or failed to give up after doing the research because it makes such a great presentation for investor storytime. [1] https://www.alertwildfire.org https://www.alertwildfire.org