6 ms·

Amazon, Google, Others Are Developing Private Air-Traffic Control for Drones

45 points by eplanit 9y ago

londons_explore 9y ago
This is a really good idea. I'd like to see some kind of REST api saying "Can I enter this X,Y,Z bounding box?", and then when getting back a HTTP 200 OK, a followup request can then be sent saying "I'm now out of this other XYZ bounding box". As the skies get crowded, craft could even "rent" their airspace by the cubic meter. If you're a craft which can guarantee to be in a smaller space and fly close to other craft, like formation flying, you use up less airspace and save money. Something like this for big planes would save massive amounts of fuel, because algorithms could be used to match flight patterns with weather maps to get a tailwind or uplift, rather than having a pilot having to manually ask ATC over the radio and then being forced to fly in a straight line at a fixed altitude as most planes do today.
- xg15 9y ago
  It might also cost large amounts of fuel when planes would have to take completely nonsensical routes to go around a scatter of tiny, blocked airspace areas everywhere. Additionally, people might start to "squat" the airspace around airports and along preferrable routes and "rent" them to airlines at an increased price. (Or, if they are a competitor or have other reasons to damage the airline, simply block it and don't rent it at all)
  - londons_explore 9y ago
    I feel like an economist could solve this problem... The "tiny blocked airspaces everywhere" problem is really how do you design an efficient market. Perhaps for example, just because you have a piece of airspace doesn't give you the right to keep it. Someone else can come along and outbid you for it auction-style. That way, anyone squatting will have to pay handsomely to do so. Obviously all the routing and bidding and allocation is done automatically and the pilot is just given a route to fly. While there is no shortage of airspace, most bids would probably be fractions of a cent anyway.
    - jonlucc 9y ago
      > Someone else can come along and outbid you for it auction-style. There's someone who registered a new TLD that was aimed at making wedding websites, and they make it so the cost increases over time to avoid squatting. I think their rates are something like <$50 for 1 year, then >$1000 for the next, but the ramp could be however steep you want and on whatever time scale.
    - xg15 9y ago
      > * Someone else can come along and outbid you for it auction-style.* Maybe I misunderstand your model, but which entity would run the auction? And who gets to collect the winning bids? In the previous post, I assumed the whole market was to be a thing between the airspace participants while the air traffic authority just keep track who has blocked which part of the airspace. Would the authority now also conduct the auctions? > That way, anyone squatting will have to pay handsomely to do so. I'm not an economist, but I don't see why the squatters couldn't have deep pockets. Squatting for political/strategic reasons might be done by airlines themselves - who might have large enough war chests to play games. (If the goal is to keep certain competitors out, they might not even need to win the auctions - driving up the price enough so that continued bidding becomes unviable would be enough) If squatting for profit is viable, there might appear speculators who use credit/investor money to stay in the aucton. > While there is no shortage of airspace, most bids would probably be fractions of a cent anyway. My suspicion is that, just like land on the ground, different areas of airspace would have vastly different value. For drone/airtaxi/etc participants, airspace above big cities would probably be very valuable - while long-distance flights need to choose the best route, while taking distance across the globe, jetstreams, weather, distance to nearest airports, etc into account.
    - UncleEntity 9y ago
      I'll take a shot at this since I'm bored... Have occupancy be based on physical occupancy with a time limit based on a reasonable value corresponding to expected airspeed for the location. Of course airports would still control the landing/takeoff order because that's what they do. I suspect the FAA has rules against blimps and helicopters hanging out in the paths the jumbojets take so I don't really see any way it would be possible to squat in a particular piece of sky. For smaller drones, I really doubt there will ever be more drones flying around than birds and they seem quite able to share the airspace with a minimum of problems aside from becoming lunch for malicious actors or colliding with ground based vehicles.
  - craftyguy 9y ago
    And before long, the FCC will be trying to assert jurisdiction over the whole situation.
  - mkempe 9y ago
    You should trust free markets and the widely distributed, deep wisdom of contracts. There are already market-based mechanisms in use for such needs in other areas of life. For example, one can rent an indoor soccer field at a fixed weekly time for a fixed annual amount, and nobody else gets to use that field during that time. Then once a year that contract can be re-negotiated if someone else is offering a better deal -- better field? higher or lower price? nicer changing room? shorter- or longer-term agreement? etc.
    - mkempe 9y ago
      If you reject free markets, contracts, and the social harmony of pricing mechanisms, you had better have better arguments than clicking on a down arrow.
    - xg15 9y ago
      I'm not an expert in air-traffic control, but I'd assume that is kind of how the current system works. After all, we do have private airlines who in some way do coordinate access to a shared resource via the air traffic authority. The innovation the parent was suggesting was that airspace is allocated/traded in realtime without any long-term arrangements in effect. (If I understood it correctly)
- cjrp 9y ago
  > Something like this for big planes would save massive amounts of fuel, because algorithms could be used to match flight patterns with weather maps to get a tailwind or uplift Flight planners already do this manually for airlines, when creating flight plans
  - iamdave 9y ago
    Yup. My housemate works for Southwest Airlines as a Flight Dispatcher, and evidently it's a career path that's in need of workers. https://www.sheffield.com/articles/aircraft-dispatcher https://www.sheffield.com/articles/aircraft-dispatcher
    - UncleEntity 9y ago
      Or, conversely, an AI based startup idea...
- karthikb 9y ago
  It's a bit more complex than this - since unlike a ground vehicle, many vehicles (especially those which are flying longer distances) can't just stop in their bounding box until the next one is free. And regardless of the vehicle type, none can wait indefinitely. Thus, you need confidence that the entire path - in 4D - is clear before setting out. I'm actually running an effort at Airbus' A^3 division to study this exact question - how to architect new air traffic systems to (safely!) enable extremely high density operations of drones, UAVs, autonomous air taxis, etc. And we're hiring. Project info: https://www.airport-technology.com/features/airbuss-foray-air-traffic-management-future/ https://www.airport-technology.com/features/airbuss-foray-ai... Job listings: https://www.airbus-sv.com/jobs/ https://www.airbus-sv.com/jobs/ , under Altiscope
neurotech1 9y ago
Non-paywall archive version: http://archive.is/76nw0 http://archive.is/76nw0
mooneater 9y ago
> talk to each other through web-based applications using web for this seems a bit heavyweight
- ocdtrekkie 9y ago
  I would characterize this as a "when you're a hammer, everything looks like a nail" thing. If you ask web companies how to have some devices communicate, it's pretty likely they're going to come up with a web platform.
  - platz 9y ago
    There are loads more web enabled devs than protocol hackers. It's a way of doing this on the cheap.
    - eftychis 9y ago
      They are getting into an area of soft real time systems with a "donkey", in my opinion. Either they are not thinking ahead or are being too optimistic or they are going to use a bunch of hardware to make this work. My thoughts. What do you think?
      - londons_explore 9y ago
        The realtime-ness of airtraffic control is really very soft... We're comparing to humans on radios here... HTTP is waaaay faster... Also, web services are well understood in terms of scalability, uptime even in the face of replacing/failing hardware, ability to shift load to other datacenters, etc. By building on HTTP, you get authentication (OAuth) and encryption (HTTPS) for free. Build a JSON webservice and you get a nice extensible protocol too, so can add new features on an almost weekly basis without disrupting existing users. By building on TCP/IP, you can allow any form of transport - satellite, wifi, bluetooth, LTE, etc. without any extra testing or certification. Most existing airtraffic control systems wouldn't be able to handle a nuke... The internet is designed to resiliently route around such things...
  - 9y ago
anonu 9y ago
I wonder if the current outdated air traffic control system will eventually be folded into this new system being built.
- iamdave 9y ago
  Wat is it about the current system that makes it outdated, and what would 'the future' (for lacking a better descriptor) look like to you? Not here to be the voice of dissent for its own sake, merely an aviation enthusiast.
- oh_sigh 9y ago
  What part is outdated? What would you improve in the current system?
NelsonMinar 9y ago
One precedent for this is FLARM, the air traffic system designed by glider pilots in Europe. It was one of the first systems deployed where aircraft had GPS equipment broadcasting their positions to each other. Developed privately and first deployed voluntarily. https://en.wikipedia.org/wiki/FLARM https://en.wikipedia.org/wiki/FLARM

45 points by eplanit 9y ago

This is a really good idea. I'd like to see some kind of REST api saying "Can I enter this X,Y,Z bounding box?", and then when getting back a HTTP 200 OK, a followup request can then be sent saying "I'm now out of this other XYZ bounding box". As the skies get crowded, craft could even "rent" their airspace by the cubic meter. If you're a craft which can guarantee to be in a smaller space and fly close to other craft, like formation flying, you use up less airspace and save money. Something like this for big planes would save massive amounts of fuel, because algorithms could be used to match flight patterns with weather maps to get a tailwind or uplift, rather than having a pilot having to manually ask ATC over the radio and then being forced to fly in a straight line at a fixed altitude as most planes do today.

xg15 9y ago
It might also cost large amounts of fuel when planes would have to take completely nonsensical routes to go around a scatter of tiny, blocked airspace areas everywhere. Additionally, people might start to "squat" the airspace around airports and along preferrable routes and "rent" them to airlines at an increased price. (Or, if they are a competitor or have other reasons to damage the airline, simply block it and don't rent it at all)
- londons_explore 9y ago
  I feel like an economist could solve this problem... The "tiny blocked airspaces everywhere" problem is really how do you design an efficient market. Perhaps for example, just because you have a piece of airspace doesn't give you the right to keep it. Someone else can come along and outbid you for it auction-style. That way, anyone squatting will have to pay handsomely to do so. Obviously all the routing and bidding and allocation is done automatically and the pilot is just given a route to fly. While there is no shortage of airspace, most bids would probably be fractions of a cent anyway.
  - jonlucc 9y ago
    > Someone else can come along and outbid you for it auction-style. There's someone who registered a new TLD that was aimed at making wedding websites, and they make it so the cost increases over time to avoid squatting. I think their rates are something like <$50 for 1 year, then >$1000 for the next, but the ramp could be however steep you want and on whatever time scale.
  - xg15 9y ago
    > * Someone else can come along and outbid you for it auction-style.* Maybe I misunderstand your model, but which entity would run the auction? And who gets to collect the winning bids? In the previous post, I assumed the whole market was to be a thing between the airspace participants while the air traffic authority just keep track who has blocked which part of the airspace. Would the authority now also conduct the auctions? > That way, anyone squatting will have to pay handsomely to do so. I'm not an economist, but I don't see why the squatters couldn't have deep pockets. Squatting for political/strategic reasons might be done by airlines themselves - who might have large enough war chests to play games. (If the goal is to keep certain competitors out, they might not even need to win the auctions - driving up the price enough so that continued bidding becomes unviable would be enough) If squatting for profit is viable, there might appear speculators who use credit/investor money to stay in the aucton. > While there is no shortage of airspace, most bids would probably be fractions of a cent anyway. My suspicion is that, just like land on the ground, different areas of airspace would have vastly different value. For drone/airtaxi/etc participants, airspace above big cities would probably be very valuable - while long-distance flights need to choose the best route, while taking distance across the globe, jetstreams, weather, distance to nearest airports, etc into account.
  - UncleEntity 9y ago
    I'll take a shot at this since I'm bored... Have occupancy be based on physical occupancy with a time limit based on a reasonable value corresponding to expected airspeed for the location. Of course airports would still control the landing/takeoff order because that's what they do. I suspect the FAA has rules against blimps and helicopters hanging out in the paths the jumbojets take so I don't really see any way it would be possible to squat in a particular piece of sky. For smaller drones, I really doubt there will ever be more drones flying around than birds and they seem quite able to share the airspace with a minimum of problems aside from becoming lunch for malicious actors or colliding with ground based vehicles.
- craftyguy 9y ago
  And before long, the FCC will be trying to assert jurisdiction over the whole situation.
- mkempe 9y ago
  You should trust free markets and the widely distributed, deep wisdom of contracts. There are already market-based mechanisms in use for such needs in other areas of life. For example, one can rent an indoor soccer field at a fixed weekly time for a fixed annual amount, and nobody else gets to use that field during that time. Then once a year that contract can be re-negotiated if someone else is offering a better deal -- better field? higher or lower price? nicer changing room? shorter- or longer-term agreement? etc.
  - mkempe 9y ago
    If you reject free markets, contracts, and the social harmony of pricing mechanisms, you had better have better arguments than clicking on a down arrow.
  - xg15 9y ago
    I'm not an expert in air-traffic control, but I'd assume that is kind of how the current system works. After all, we do have private airlines who in some way do coordinate access to a shared resource via the air traffic authority. The innovation the parent was suggesting was that airspace is allocated/traded in realtime without any long-term arrangements in effect. (If I understood it correctly)
cjrp 9y ago
> Something like this for big planes would save massive amounts of fuel, because algorithms could be used to match flight patterns with weather maps to get a tailwind or uplift Flight planners already do this manually for airlines, when creating flight plans
- iamdave 9y ago
  Yup. My housemate works for Southwest Airlines as a Flight Dispatcher, and evidently it's a career path that's in need of workers. https://www.sheffield.com/articles/aircraft-dispatcher https://www.sheffield.com/articles/aircraft-dispatcher
  - UncleEntity 9y ago
    Or, conversely, an AI based startup idea...
karthikb 9y ago
It's a bit more complex than this - since unlike a ground vehicle, many vehicles (especially those which are flying longer distances) can't just stop in their bounding box until the next one is free. And regardless of the vehicle type, none can wait indefinitely. Thus, you need confidence that the entire path - in 4D - is clear before setting out. I'm actually running an effort at Airbus' A^3 division to study this exact question - how to architect new air traffic systems to (safely!) enable extremely high density operations of drones, UAVs, autonomous air taxis, etc. And we're hiring. Project info: https://www.airport-technology.com/features/airbuss-foray-air-traffic-management-future/ https://www.airport-technology.com/features/airbuss-foray-ai... Job listings: https://www.airbus-sv.com/jobs/ https://www.airbus-sv.com/jobs/ , under Altiscope

neurotech1 9y ago

Non-paywall archive version: http://archive.is/76nw0 http://archive.is/76nw0

mooneater 9y ago

> talk to each other through web-based applications using web for this seems a bit heavyweight

ocdtrekkie 9y ago
I would characterize this as a "when you're a hammer, everything looks like a nail" thing. If you ask web companies how to have some devices communicate, it's pretty likely they're going to come up with a web platform.
- platz 9y ago
  There are loads more web enabled devs than protocol hackers. It's a way of doing this on the cheap.
  - eftychis 9y ago
    They are getting into an area of soft real time systems with a "donkey", in my opinion. Either they are not thinking ahead or are being too optimistic or they are going to use a bunch of hardware to make this work. My thoughts. What do you think?
    - londons_explore 9y ago
      The realtime-ness of airtraffic control is really very soft... We're comparing to humans on radios here... HTTP is waaaay faster... Also, web services are well understood in terms of scalability, uptime even in the face of replacing/failing hardware, ability to shift load to other datacenters, etc. By building on HTTP, you get authentication (OAuth) and encryption (HTTPS) for free. Build a JSON webservice and you get a nice extensible protocol too, so can add new features on an almost weekly basis without disrupting existing users. By building on TCP/IP, you can allow any form of transport - satellite, wifi, bluetooth, LTE, etc. without any extra testing or certification. Most existing airtraffic control systems wouldn't be able to handle a nuke... The internet is designed to resiliently route around such things...
- 9y ago

anonu 9y ago

I wonder if the current outdated air traffic control system will eventually be folded into this new system being built.

iamdave 9y ago
Wat is it about the current system that makes it outdated, and what would 'the future' (for lacking a better descriptor) look like to you? Not here to be the voice of dissent for its own sake, merely an aviation enthusiast.
oh_sigh 9y ago
What part is outdated? What would you improve in the current system?

NelsonMinar 9y ago

One precedent for this is FLARM, the air traffic system designed by glider pilots in Europe. It was one of the first systems deployed where aircraft had GPS equipment broadcasting their positions to each other. Developed privately and first deployed voluntarily. https://en.wikipedia.org/wiki/FLARM https://en.wikipedia.org/wiki/FLARM