26
$\begingroup$

The US Military's use of drones has become commonplace, and widely known at this point. Although these vehicles are operated from nearby bases during these attacks, it seems like the latency (time of transmission) would be problematic as far as flying goes. Are these vehicles typically flown 100% by autopilot, where there's very little real-time reaction?

$\endgroup$
10
  • 2
    $\begingroup$ What latency? Radio waves can travel around the earth about 7.5 times per second. Let's imagine your drone is 10,000 miles away - a long way - a radio signal can travel there and back about 9.5 times per second. $\endgroup$
    – Simon
    Sep 29, 2015 at 11:30
  • 14
    $\begingroup$ @Simon so, that's a 100ms lag. talk to any gamer and tell them that's not a huge latency. (of course you are not factoring the lag added by the various transmission equipement.) $\endgroup$
    – njzk2
    Sep 29, 2015 at 16:27
  • 10
    $\begingroup$ @Simon: Online gamers, especially shooters, have to deal with lag when controlling the game. Their controls are quite similar on a broad level to drone controlling, as there's essentially a feedback loop separated by connection lag from the actual system. $\endgroup$ Sep 29, 2015 at 16:44
  • 3
    $\begingroup$ @Simon: For gamers, the latency is not the result of the distance but of the delay in the equipments along the packets path (routers, amplifiers). Companies like Akamai provide low latency routing (for a price). For a drone the latency would come from the distance of the satellite(s) relaying the commands. $\endgroup$
    – mins
    Sep 29, 2015 at 18:01
  • 7
    $\begingroup$ 100 ms lag is probably negligible in this case. this isn't an FPS shooter. Go fly a cessna and observe the latency between providing a control input and the airplane doing anything and you'll find that is greater than 100 ms. As a pilot you learn to anticipate and that is how you overcome lag. $\endgroup$
    – casey
    Sep 29, 2015 at 18:04

4 Answers 4

25
$\begingroup$

There are multiple ways to fly and control Unmanned Aerial Vehicles (UAV), or drones, for e.g. the Predator.

Predator System

Source: science.howstuffworks.com

  • Some UAVs (like the Global Hawk) can perform the entire mission (from takeoff to landing) autonomously, negating the need for a pilot.
  • Most of the UAVs, like Predator can perform (i.e. fly in) some simple missions (like reconnaissance) autonomously.
  • In case a geostationary satellite is used for communication with the UAV, the latency is around 300 ms (The electromagnetic waves have to travel ~70000 km). In case of other delays (like equipment etc.) this will not be greater than 500 ms or around half a second. This is not significant in case of slow moving systems like Predator (they are not dog-fighting, after all) and the 'pilot' can effectively control the aircraft.
  • The only situation where this is a problem is when the UAV is taking off/landing and in this case, the control is handed over to local LoS operators.

The Predator is reported to have a 'latency' of around two seconds, which causes problems during takeoff/landing. According to the telegraph,

But the two-second delay between a pilot moving a joystick in Nevada and an aircraft responding in Afghanistan is enough to cause a crash during take-off and landing. Crews in Afghanistan control 'launch and recovery’ through direct contact with antennae on the aircraft. Half an hour after take-off, control of the Reaper is handed to a crew in Nevada; half an hour before landing, it returns to the crews on the ground in Kandahar.

Note that the RAF also does things the same way. According to the report on formation of 'drone' squadron' at RAF Waddington,

Although the personnel will be permanently based at RAF Waddington, some will go to Afghanistan to control the take-off and landing of the drones locally.

$\endgroup$
17
  • 3
    $\begingroup$ I think your latency figure is one-way, but unfortunately, because it takes time for the updated instrument signal to travel from the drone to the pilot, and time for the control signal to travel back to the drone, you have to measure it two-way, round-trip. $\endgroup$ Sep 29, 2015 at 16:47
  • 6
    $\begingroup$ @TomMcW It's cheaper to have one guy in Afghanistan launch a couple of airplanes and a bunch of pilots in Nevada than ship the bunch of pilots to Afghanistan - after some distance, communication is going via satellite either way. Besides, the guy in Afghanistan has a risk of being hit by things that go boom, the guy in Nevada not so much. $\endgroup$
    – Sanchises
    Sep 29, 2015 at 16:59
  • 3
    $\begingroup$ @TomMcW: There is little to no relation between what a Predator drone does in real life, and what you do with your little first person shooter computer games in your bedroom. $\endgroup$ Sep 29, 2015 at 23:21
  • 6
    $\begingroup$ @TomMcW Get satellite internet and you'll find your games unplayable. If we had the luxury of laying millions of miles of wiring all over the potential battlefronts around the world so we could then erect antennae all over to maintain constant line-of-sight contact with the drones, the amount of latency would be comparable to using the internet to play games. But trust me, I've tried, and you just don't get good latency on satellite connections (although you could get really awesome bandwidth). $\endgroup$
    – tpg2114
    Sep 30, 2015 at 2:29
  • 5
    $\begingroup$ @TomMcW It wasn't Microsoft, but it was a Bermuda based company that ran a cable from the US to Australia. Satellite communication makes for long latency times because geosynchronous orbit is pretty far away and the signal has to go out and back before it gets to you (plus all kinds of cables, hardware, and software in between). 8 times slower is not that hard to believe. $\endgroup$
    – tpg2114
    Sep 30, 2015 at 3:42
7
$\begingroup$

The aerodynamic surfaces of a drone are controlled by computers, not by humans. When a human makes a control input this input is broadcast to the UAV, and the computers on the UAV make the appropriate changes to the flight surfaces to carry that input out. It's not direct control of the drone by humans via radio.

$\endgroup$
5
  • $\begingroup$ Do you mean the drone has some level of autonomy and can react to events without the immediate intervention of the pilot? Maybe the pilot only send waypoints / targets, and the drone manages to go there? $\endgroup$
    – mins
    Sep 29, 2015 at 18:08
  • 5
    $\begingroup$ I think what he's saying is that the drone operator will move the joystick, and that tells the plane "Bank right 10 degrees" and the avionics computer figures out how to do that. Just like a fly-by-wire fighter jet would do. It's not like an hobbyist R/C plane where moving the stick directly controls the flight surfaces. $\endgroup$
    – JPhi1618
    Sep 29, 2015 at 19:22
  • $\begingroup$ @JPhi1618: That's what I was thinking too, just like entering commands on the MCP of a B737. $\endgroup$
    – mins
    Sep 30, 2015 at 0:03
  • $\begingroup$ That's exactly what I mean @JPhi1618 $\endgroup$
    – GdD
    Sep 30, 2015 at 7:44
  • $\begingroup$ "Do you mean the drone has some level of autonomy and can react to events without the immediate intervention of the pilot? " Isn't that the definition of Drone? Otherwise it would just be a big R/C plane. $\endgroup$
    – roel
    Sep 30, 2015 at 10:04
2
$\begingroup$

Source: Acquaintances who worked on avionic software and hardware development

  1. Latency matters for anything mechanical that flies !
  2. On-board equipments are developed to handle sub-second interactions. Any interaction from ground is assumed to take many 100s of milliseconds to reach the flight (anything that flies). Humans operating the flight are trained to be quick and anticipate rather than react !
  3. To minimize the impact of latency, a series of information keeps flowing thereby converting latency problem into throughput problem. Throughput problems can be handled using multiple beacons / transmitting sources. This does not reduce latency but pre-loads onboard systems with accurate data to anticipate next 500 mS or so.
  4. To understand the solutions for latency, look for 2 businesses that are latency sensitive - gaming and high frequency trading. To put things in perspective, in HFT, a market event is detected by software and it generates an order and sends it - all under 1 micro-sec. Further latencies in network stack is not measured here but it gives you a sense of what operates here.
$\endgroup$
3
  • 1
    $\begingroup$ In the case of high-frequency trading, it seems that what's really going on is not that companies wish to react to e.g. the fact that a stock's price drops, but rather that they want to be able to have a limit buy order pending in such a fashion as to be invisible until such time as it could be fulfilled. The same sort of semantics could be offered without need for instant communication if the market had a trustworthy entity which would pair up buyers and sellers without revealing any information about pending limit orders. In any case, the key to efficiency is to express decisions... $\endgroup$
    – supercat
    Sep 29, 2015 at 19:07
  • $\begingroup$ ...made on the basis of incomplete information, in a way that indicates in advance what should be done with information that becomes available but which wasn't available when the decision was made. $\endgroup$
    – supercat
    Sep 29, 2015 at 19:10
  • $\begingroup$ I think everything you said except about HFT. HFT avoids latency in the host, but they also do everything to shrink the transit time, whereas drones cannot really do that. $\endgroup$ Sep 30, 2015 at 3:41
0
$\begingroup$

A significant number of these devices use "top/level" control. The operator may have a camera to see what is in front of the device but the main screen they are looking at is a top view of the landscape (including height map) of where it is flying. The altitude is set to x meters/yards above ground/sea level. The device maintains height and direction based on this input. As the drone isn't in live action and dodging anything - the delays and adjustments do not impact the route of the machine.

E.g. A 90 degree turn only requires 2 signals. One to initiate the turn and the other back to base to confirm the turn is complete and the drone's current positioning.

You would rarely find a full scale drone needing manual adjustment of anything less than 500 meters. Also the video footage from the drone is usually only uploaded at full quality once it has returned home.

$\endgroup$
1
  • $\begingroup$ Welcome. You may want to edit your question (using the edit link) and include your comment content and delete it. $\endgroup$
    – mins
    Sep 30, 2015 at 5:32

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .