Autonomous Drones May Be the Future, but Israeli Army Still Stresses Human Factor

A visit to the Palmahim air force base shows that the human element in operating unmanned aerial vehicles will remain for the foreseeable future.

An Israeli-made Haron drone on a runway in the Palmachim airbase.
IDF Spokesperson's Unit

Autonomously-driven vehicles are progressing in the civilian world at a rapid pace in recent years, despite various technical problems and failures alongside bureaucratic hurdles. Google’s self-driving cars have racked up millions of miles on the roads, almost without accidents. So far only one reported death has been linked to such vehicles, in a semi-autonomous Tesla.

A recent report from the U.S. Defense Department called to “accelerate [the department's] exploitation of autonomy — both to realize the potential military value and to remain ahead of adversaries who also will exploit its operational benefits.” For years, experts have been saying that the new F-35 airplane belongs to the last generation of manned combat aircraft. Future generations are expected to be autonomous and remotely controlled.

However, whether it's for precision or public relations, the Israel Air Force has followed the lead of the U.S. Air Force in deciding to remind people that unmanned aerial vehicles are not robots acting independently. The human factor is always part of the decision-making loop, and it seems that it will remain there for the foreseeable future. At least this is the conclusion this reporter drew from a visit of technology correspondents to the IDF’s remotely piloted aerial vehicle (RPAV) squadrons and training facilities at the Palmahim air force base.

The promises of artificial intelligence and autonomous objects still seem a long way away from the control vans at Palmahim. When asked about how critical the human factor was to the operation, a senior officer said: “100%.”

“Forget about the movies. This is not a kid’s model aircraft, it is an operational tool, and it is no different than any other tool,” he added.

When the base commander, Brig. Gen. N. (whose name cannot be published for security reasons), was asked about these predictions about future aircraft, he was not ready to give up so quickly on human involvement. “We always look at the operational challenge. True, there is a machine, but the question is, what does it do for you? In the end, the Air Force has a mission and in decision-making, I want to hope there will always be a human,” he said.

Daniel, the commander of a squadron of Hermes 450 drones, is of the same opinion. “An RPAV is the people who prepare it for flight and the people who fly it. We must choose where the person is critical and where I trust the technology,” he said. The squadron is making progress toward automating take-off and landing, which is now controlled by operators outside the control vans — though the core operations will still remain in human hands.

The human touch, or operators in this case, are mostly young people who completed at least a year of pilot training before leaving the course. They learn to fly drones over a six-month period and spend three months in advanced training. Only after that can they qualify for operational duty.

An operator in an Israel Air Force base pilots an Israeli Heron (Eitan) drone, April 5, 2011. A man in an army uniform sits in front of a bank of four monitors showing coordinates and a map, one hand on a specialized computer keyboard and the other on a joystick.
Olivier Fitoussi

The IDF’s main drone platforms are the Hermes 450, made by Elbit Systems, which entered service back in 1999; the  Heron TP (Eitan), made by Israel Aerospace Industries; Elbit’s Hermes 900 (Kochav); and the Heron (Shoval).

All these drones can spend at least 20 hours in the air at a time. The operators work in four-hour shifts, though they can choose to work eight- or nine-hour shifts instead. Today, about 65 percent of drone flight time is operational, but the Air Force hopes to raise this number to almost 90 percent by increasing investments and shifting more training hours to simulators.

The simulators allow much greater flexibility for various training scenarios, at a lower cost. (Throughout the visit, the budgetary advantage of using drones came up time and time again.)

The Pentagon report, as well as the IDF, both note the major advantage in safeguarding human lives. Drones account for 70 percent of all operational Air Force flight time, including intelligence gathering.

The technological leap of the past few decades has changed direction to a great extent: If in the past we assumed new technologies were born in the military (especially during wartime) as well as in government agencies such as NASA and DARPA in the United States, and the Israeli defense ministry’s weapons and technological development administration, in recent years the civilian world has invented advanced off-the-shelf technologies at a pace that government bodies find hard to keep up with. This has led to the adoption of open-source platforms and hackathons within the IDF, but during our visit to Palmahim, it was clear that open-source code had a marginal role in the development of what we saw.

The senior Air Force officers emphasized the cooperation between the local defense industries and the military. Many of the drone operators go to work in the industry after they finish their service, which helps keep good relations between both organizations. It also speeds up development cycles for features such as special payloads or day and night vision packages, which not only improve object identification on the ground but also helps prevent misidentification, as happened during Operation Cast Lead in Gaza when a truck was bombed after the cooking gas canisters it was carrying were misidentified as rockets. One officer said that it is now sometimes possible to tell whether the person on the screen is a man, woman or child, and even the individual's emotional state.

The applications involved also undergo a rapid development processes. The flight control applications are updated every few months, and are becoming more and more like the navigation apps we use on our mobile phones. It's possible to control the drone’s flight path with just a few touches of the screen.

But it's not yet Waze, as we realized from our brief experience. Even the basic control functions for the cameras over urban areas have a steep learning curve. When we tried keeping the camera over a specific route, we found it was not at all trivial. An officers who was asked about the possibility of marking the planned route in order to help the operator said the development of such tools was still more complicated than developing features for civilian apps. Here, there is no option for beta trials, it must work as perfectly as possible — even if problems do occur, and always will.

“The battlefield is always changing, we always have our eye on development and try to be a bit in front of it,” said Daniel. And as often happens with the military, many of our questions were left unanswered.