Last year the large U.S. Air Force UAVs (MQ-1 Predator, MQ-9 Reaper, and RQ-4 Global Hawk) had an accident rate of 3.8 per 100,000 hours. This accident rate is currently lower than that of the F-16, which is currently one of the safest manned fighter aircraft flying. Last year these air force UAVs had 13 "Class A" accidents (one causing over two million dollars of loss). By way of comparison, the F-22 an accident rate is about 6 per 100,000 hours. F-15s and F-16s have an accident rate of 4 per 100,000 flight hours.
Despite the current accident record, it’s been a rough decade for air force UAVs. Some 20 percent of the air force Predator and Reaper UAVs have been lost to accidents. This spurred the air force to make UAVs more reliable and reduce the loss rate. Two years ago the accident rate for its MQ-1 Predators was down to about 5 and it was expected that this would continue declining. The year before that the UAV rate was twice the rate of manned fighter aircraft (like the F-15 or F-16) and four times the rate of the old but very reliable B-52. Note that the UAV accident rate is lower than that of single engine private aircraft (8). Reapers have a slightly higher rate than the older (and more numerous) Predator.
Only a few years ago the loss rate for the 1.1 ton MQ-1 Predator was 30. The 4.7 ton larger MQ-9 Reaper had a loss rate of about 15 two years ago, after four years in service. It was a decade ago that the MQ-9 made its first flight. The Predator has been in action since the late 1990s. The design and operation of the MQ-9 learned much from the experience of the MQ-1. For two decades now engineers have been seeking ways to make these unmanned aircraft more reliable and resistant to accidental loss. It’s been difficult because these aircraft have been flown frequently, in many different climates and under combat conditions for the past decade. Safety features that worked in Iraq often did not work in Afghanistan. UAV operator training had to be tweaked as well. Experienced pilots of manned aircraft often did less well operating UAVs than non-pilots who learned how to fly on simulators. It seems that there were things pilots of manned aircraft had to unlearn before they could handle UAVs more effectively.
There was never any doubt that the accident rates would come down, and many engineers felt it was possible to make UAVs safer than manned aircraft. Previously, unmanned aircraft always had a much higher loss rate, which was partly the result of not having a pilot on board (who could “feel” the aircraft) and the inability to compensate for that. Older model UAVs had much higher rates. The 1980s era RQ-2A Pioneer had an annual rate of up to 363 per 100,000 hours. Despite that, the RQ-2 proved very useful during the 1991 Gulf war.
Despite UAVs being cheaper, the air force always knew they would have to reduce the accident rate to make these aircraft competitive with manned ones. This was ultimately accomplished with cheaper and more effective sensors and flight controls, as well as better analysis of how operators succeeded or failed.