Somewhere over Kabul


Figure 1. A very near miss between a German Army Luna UAV and an Airbus A300


Sometimes a picture tells a thousand words

The picture above is the penultimate image sent from an German army UAV after it almost became a participant in the first UAV and commercial aircraft mid-air collision. The next image shows the UAV on its way towards the ground having been knocked into an unrecoverable upset by the aircraft’s wake, a very close call indeed.

Now I’d assume that mixing it up with commercial aircraft was not high on the list of potential operational scenarios considered when developing this UAV, and that this assumption went on to shape the UAV design. Of course if we then go on to operate the UAV system in a low intensity conflict where military and commercial flights coexist and where ATC is rudimentary at best the potential for a classic ‘interaction’ accident is introduced.

This neatly illustrates the problem we face using old systems in new context where it’s not the hazards we can predict that bite us but the the unknowns introduced by operating a system outside the intended operational context.

Similarly when we field a novel system for the first time it introduces a new set of “unknown unknown’s” that traditionally are discovered the hard way.

As an example of such unknown, unknowns a new ‘air safety’ hazard introduced by UAV technology one might somewhat naively assume that a UAV’s control station is manned all the time, much in the same way that an aircrafts cockpit in flight is continuously manned. But unlike an aircraft there is nothing to stop an operator from walking out to answer a call of nature, answer a  mobile phone message or grab a cup of coffee.

So there is the real potential for the UAV to completely unmanned, unlike a traditional manned aircraft where the pilots are always in the aircraft (at least). This may not be such a likely event in a highly disciplined and structured organisation (e.g the military). But as UAVs become more pervasive, cheaper, smarter (and their operators less skilled) will this always be true?

Figure 2. A German Army (Luna) UAV control station illustrating the significant differences between piloting an RPV and a traditional manned aircraft.

Looking ahead how will we manage the possibility of fully autonomous vehicles that are waiting in the wings, current air traffic control is based (largely) upon voice communication, so will this be workable in the future?

How will air traffic control deal with high flying ‘micro’ UAVs, which though small could still do significant damage if sucked into an commercial jet’s engine for example. And similarly what of the concept of ‘swarm’ UAVs which autonomously interact and demonstrate emergent behaviour? What about hacking or jamming of RPV data links?

Considering for a moment changes to the operational context again, how will the FAA’s ‘free flight’ concept interact with unmanned and/or fully autonomous air vehicles in this regard? For example if path deconfliction is used to prevent collision this pre-supposes that paths (and stay out areas) can be defined ‘a priori’ during mission planning. If a full ‘free flight’ capability is implemented then will we require automation or remote operators to make flight critical decisions in a less structured and more complex environment?

Imagination and creativity is I think a necessary prerequisite for identifying new hazards and dealing with them, preferably before they result in a smoking hole in the ground…