Human Factors of UAVs: Manning the Unmanned

Transcription

1 Human Factors of UAVs: Manning the Unmanned

2 Opening Session Overview Welcome ( ) Opening Remarks ( ) The Operator s Perspective ( )

3 Welcome! Introductions Logistics

4 Opening Remarks Human Factors of UAVS: Mishaps and Myths

5 UAV Mishaps High mishap rates are frequently cited as a deterrent to more widespread adoption of UAVs into the force structure. Inside the Pentagon, May 6, 2004 Currently, the UAV accident rate is 100 times higher than that of manned aircraft. Paul Jackson, Jane s All the World s Aircraft , (Alexandria, VA: Janes Information Group, 2003), pp

6 UAV Mishaps Pioneer has unacceptable Class A mishap rate: 385 mishaps per 100,000 flight hours (counted since 1986). In contrast, the manned Naval Aviation mishap rate is 2 per 100,000 flight hours Naval mishap*: unplanned event or series of events directly involving naval aircraft, which results in $10,000 or greater cumulative damage to naval aircraft or personal injury. Class A mishaps: that total amount of damage exceeds $1,000,000 or if the air vehicle is destroyed.

7 UAV Reliability Comparison Unmanned Aircraft 1 Accidents/100,000 hr Global Hawk (4 accidents)* 168 RQ-2A Pioneer 363 RQ-2B Pioneer 139 Predator RQ-1A* 43 Predator RQ-1B* 31 Manned Military Aircraft 1 F Manned Civil/ Commercial Aircraft Accident Rates (2003) 2 General Aviation 6.7 General Aviation (Fatal) 1.4 Part 121 Scheduled & Unscheduled Part 121 Scheduled & Unscheduled (Fatal) Global Hawk Pioneer From R. John Hansman rjhans@mit.edu Sources: 1.) National Defense Magazine, May 2003; 2.) NTSB Press Release, March 22, 2004 Predator B *Total Operational Hours < 100,000 : Predator (65,000 hr); Global Hawk (2,500 hr)

8 UAV Reliability Trends 2000 Cumulative Mishap Rate per 100,000 Hrs RQ-1 / Predator RQ-2 / Pioneer RQ-5 / Hunter Year From R. John Hansman rjhans@mit.edu Source: Feb Office of Secretary of Defense (OSD) UAV Reliability Study

9 Reliability Trends From R. John Hansman Source: Feb Office of Secretary of Defense (OSD) UAV Reliability Study

10 Mishap Rate vs. Cost From R. John Hansman

11 Mishap Causes Schmidt & Parker (1995) of the Naval Safety Center examined 107 mishaps that occurred between 1986 & % of mishaps occur as a result of electromechanical problems 33% due to human error*, such as: crew selection & training aeromedical readiness pilot proficiency/currency personnel shortages operational tempo human error in teamwork & aircraft control Seagle (1997) analyzed 203 mishaps from 1986 through % due to human error* *Human System Failure Fergusen, M.G. (1999). Stochastic modeling of naval unmanned aerial vehicle mishaps: Assessment of potential intervention strategies, Naval Post Graduate School Thesis

12 QF-4E Drone (FY 02) Mishap Causes Lack of feedback to controller Displays and control arrangement During a launch in ATO mode, the ATO failed to progress normally, so the primary drone controller attempted to take control manually and fly it off the runway. The primary attitude reference system, the ANI AJB-7, failed. Among other things, the ANI AJB-7 provides primary attitude references to the controller's computer screen. This allows the controller to monitor the drone's performance and take over manually when necessary. At rotation speed of 160 knots, the attitude displayed on the controller's monitor indicated a normal takeoff attitude of twelve degrees nose-high. Post-mishap analysis of telemetry data revealed the actual attitude to be four degrees less. Because of this discrepancy, the controller thought the takeoff was progressing normally when in reality the MA did not rotate enough to lift off the runway. By the time the controller recognized the failure to rotate, the MA was traveling at a speed of over 250 knots. The controller went to manual control, and due to the high airspeed, commanded an excessive nose-up pitch rate. The MA was destroyed upon impact in a heavily wooded, unpopulated area. Several human factors, including the lack of "seat of the pants" feedback to the controller and arrangement of console monitors and controls, substantially contributed to the excessive nose-up command.

13 Mishap Causes Predator (FY 02) Handoff problem Procedures error A mishap occurred in Operation ENDURING FREEDOM during a planned handoff of control between two ground control stations using a lost link recovery orbit point. When accomplishing this lost link handover, the mishap crew failed to accomplish checklist steps in the proper order, resulting in the MA's Stability Augmentation System (SAS) and engine ignition being turned off immediately after enabling the satellite command link. When the MA received the commands to turn the SAS and the engine off, it entered an uncommanded dive. The MA impacted the ground nine miles off the end of the runway.

14 Mishap Causes Situation awareness Poor communication RQ-1 Predator According to the accident investigation board report, the Predator experienced a fuel problem during its descent. Upon entering instrument meteorological conditions and experiencing aircraft icing, the Predator lost engine power. The two Predator pilots, who control the aircraft from a ground station, executed critical action procedures but were unable to land the aircraft safely. It crashed in a wooded area four miles south of. According to the report, the pilots' attention became too focused on flying the Predator in icing and weather conditions they had rarely encountered. The report also cites lack of communication between the two pilots during the flight emergency as a cause of the accident.

15 Mishap Causes Action slip Global Hawk (1999) During a March 1999 test flight of AV-2, controllers mistakenly sent a signal to terminate flight, causing the UAV to crash.

16 Mishap Causes Control problems Predator RQ-1L (12/11/2003) LANGLEY AIR FORCE BASE, Va. Operator error caused an RQ-1L Predator remotely piloted aircraft to crash Dec. 11, The pilot over-controlled the aircraft's pitch commands during a nose-high unusual attitude (or uncommanded aircraft position). The aircraft abruptly pitched upward due to a software program anomaly and the pilot overcorrected to bring it back down. The abrupt input began a cycle of upward and downward movements that caused the aircraft to exceed its operating limits and eventually crash.

17 UAV Myths

18 UAV Myths Automation Fallacies: The system is automated. The more automation the better. If a problem arises a person can step in and deal with it. Thirty years of sponsored research demonstrates that automation changes the human s task and not necessarily in a positive manner. Many mishaps attributed to human out-of-the-loop problems The true beauty of the UAV system is the ability for humans to override the automation to do dynamic re-tasking

19 UAV Myths Unmanned Fallacies: There are no humans and there is therefore no need for human factors. Unmanned does not mean uncontrolled Operators are remote relative to UAV not absent Ground personnel are numerous (1 to 5 Global Hawks requires 28 maintenance persons in theater) Cost savings on the vehicle currently exceeded by costs on the ground (ground equipment, net increase in people, training)

20 UAV Myths Air Traffic Control Fallacies: Air traffic controllers monitor dozens of vehicles, why not UAV operators? One operator per vehicle is state of the art ; 1:4 is the logical extension. The UAV control task involves much more than monitoring and control of aircraft position Dynamic re-tasking and re-planning maximally exploits the UAV system State of the Practice is 2:1 Research suggests 1:n is problematic

21 UAV Myths Another question the J-UCAS program is grappling with is how many vehicles a single operator will be able to manage at once, according to Francis. "We're revisiting the management paradigm," he said. "We're asking questions like, 'How about we have different crews for health and status, for trajectory management, and for targeting?' You know the whole management paradigm is up in the air when you don't have to map one-to-one [to a cockpit]." --Jefferson Morris Aerospace Daily & Defense Report, April 12, 2004

22 UAV Myths Manned Flight Fallacy: How can it be different from manned flight? A UAV is a vehicle. UAV piloting is the same as piloting in the cockpit. A single pilot should be sufficient. It is not a vehicle, but a system that includes vehicle(s), ground control, air operations, operator(s), intelligence, weather personnel, payload operators, maintainers Piloting analogy ignores years of studies on time lag, loss of visual cues, depth perception, etc. Piloting analogy ignores the system functions beyond flight (i.e., re-tasking, replanning, sensor operation)

23 Some Human Factors of UAVs Issues Soda straw videos Effective use of automation Spatial orientation Fatigue during lengthy missions Vigilance Number of operators Effective feedback Checklists & procedures Reachback capabilities Communication Appropriate training Crew selection Geographical displays Distributed teamwork Target detection Multiple platforms Integration with airspace

24 The Operators Perspective