This chapter begins with a discussion of the potential benefits that increasingly autonomous (IA) systems may be able to provide to civil aviation, in terms of general criteria such as safety and cost. It then addresses specific applications of IA systems for air traffic management (ATM) and for various classes of aircraft. This chapter also reviews the use of IA systems for nonaviation applications.
Estimates of the current market for commercial UAS range from $5.9 billion annually in the United States to $89 billion worldwide.1,2 The drivers behind the growth of UAS include the potential to increase safety and reliability, reduce costs, and enable new operational capabilities. However, unless IA systems are implemented in a careful and deliberate manner, the actual benefit of IA systems could be limited or they could even reduce safety and reliability or increase costs. In addition, the potential benefits that accrue from the introduction of advanced IA systems in civil aviation, the associated costs, and the unintended consequences that are likely to ensue will not fall on all stakeholders equally. In fact, some stakeholders may not benefit at all from advances that greatly benefit other existing or new stakeholders. For example, making it easier for farmers to operate unmanned aircraft over their fields may be of great economic benefit to them, but it may inconvenience general aviation pilots if they must adjust their operations to avoid conflicts with agricultural unmanned aircraft systems (UAS). In addition, it would take considerable resources for the Federal Aviation Administration (FAA) to address issues related to the certification and operation of IA systems on crewed and unmanned aircraft.
Commercial air transportation in the United States operates at unprecedented levels of safety and reliability. The fatal accident rate for U.S. commercial air carrier operations is so close to zero (literally zero fatalities for
1 Nick Wingfield and Somini Sengupa, Drones set sights on U.S. skies, New York Times, February 17, 2012, http://www.nytimes.com/2012/02/18/technology/drones-with-an-eye-on-the-public-cleared-to-fly.html.
2 Daisy Carrington and Jenny Soffel, 15 ways drones will change your life, CNN.com, November 18, 2013, http://edition.cnn.com/2013/11/03/business/meet-your-friendly-neighborhood-drones/.
2010-2012 3) that further increases in safety and reliability can only be made in very small increments. Accident rates in some other segments of civil aviation, however, are substantially higher. For example, general aviation accidents resulted in 432 fatalities in 2012,4 most of them attributed to loss of control, controlled flight into terrain, and other pilot-induced errors. Most general aviation flights have a single pilot on board, and advanced IA systems could carry out some of the duties of a copilot, such as monitoring systems and identifying potential solutions to emerging problems. However, IA systems must be relatively inexpensive if they are to improve general aviation safety. Many existing safety systems, such as the Traffic Collision Avoidance System, that are required on commercial air transports are installed on very few general aviation aircraft because of their cost.
Neither the general public nor the aviation community would tolerate any decrement in safety as a result of changes in the NAS associated with introduction of advanced IA systems. Fortunately, IA systems have the potential to increase safety because automated and autonomous systems do not become distracted, do not tire, are not affected by emotions such as fear, and will not be influenced by other outside pressures, such as the perceived need to complete a flight to some specific airport at a given time. Further, advanced IA systems can have the potential to accommodate a wide variety of dynamic conditions and to react at cyberspeeds to abrupt changes. Theoretically, properly designed IA systems might compensate for certain limitations in human performance that are associated with incidents and accidents.
In general, safety and reliability would be enhanced by IA technologies that are able to execute tasks such as the following:
- Adapt to changing patterns and preferences;
- Remain vigilant at all times;
- Increase the ability to tailor actions to specific circumstances and add flexibility to plans so that they better fit the immediate demands of the situation;
- Increase the situational awareness of human operators by presenting information in a context-sensitive fashion;
- Monitor human actions and alert and/or intervene to prevent errors from causing incidents or accidents; and
- React quickly to avoid critical situations such as a collision.
In addition, advanced IA capabilities that enable UAS to replace crewed aircraft eliminate the need for an air crew and, thus, eliminate the possibility of onboard fatalities in case of an accident. If an aircraft crashes into a populated area, casualties on the ground are likely to be far less if the accident involves a small unmanned aircraft instead of a crewed aircraft, which is certain to be much larger and carry more fuel. To fully realize these safety improvements, safety standards established for the certification and operation of UAS will need to take into account how the accident rate of UAS compares to the accident rate of the crewed aircraft that UAS would replace and of the potentially large number of UAS that would be used for applications that are not currently performed using aircraft.
IA systems have the potential to reduce costs by reducing the need for highly skilled operators, by enabling more efficient operations, and by relying on UAS to conduct missions that would otherwise be executed by crewed aircraft.
The roles of humans in the NAS will change as advanced IA systems are introduced. The need for humans to fill some existing roles may be reduced or disappear altogether, while other roles may become more important or more prevalent. In addition, some new roles may arise. Coordination among all human operators will remain essential, though such coordination will likely be assisted by and necessarily include coordination with IA systems that have taken on important roles. The changing roles of human operators will reduce costs to the extent the NAS can continue to operate with fewer highly skilled operators. In some cases, highly skilled operators such as pilots could be replaced by less-skilled personnel on the ground (UAS operators). For example, remote crew might be used to augment the onboard crew to reduce crew requirements during long-haul flights and to provide better
opportunities for crew rest. 5 However, IA systems could also increase the need for highly skilled support staff, in particular the engineers needed to develop IA systems, develop new verification, validation, and certification (VV&C) standards, and assure that they perform safely and reliably over their lifetime.
The traditional approach to dealing with situations where the cognitive demand required to complete a given set of specified tasks is beyond the cognitive capacity of a single person is to divide the tasks among multiple personnel so that no individual is overloaded. For example, in airspace with a large number of aircraft, air traffic control tasks may be divided so that aircraft in a particular airspace volume are under the control of an R-side (or radar) controller, who is responsible for monitoring the position of aircraft and exercising tactical control, and a D-side (or data) controller, who is responsible for managing the data required to make and communicate decisions while also exercising strategic control. Increased autonomy in ATM could lead to the shedding of certain tasks that are currently performed by humans. This in turn could increase the number of aircraft that are handled by each pair of R- and D-side controllers and/or (in the extreme) eliminate the need for one of the controllers. These changes could increase the efficiency with which aircraft are handled and, thus, the efficiency of NAS operations while also reducing the need to increase the number of highly skilled controllers as the level of air traffic increases.
The cost of developing, manufacturing, and operating unmanned aircraft—especially small unmanned aircraft—can be much less than the cost of crewed aircraft that would execute the same missions. Thus, in many cases costs could be reduced by replacing crewed aircraft with highly capable UAS. In other cases, UAS will enable new economic activity by enabling businesses to provide services that could not be profitably executed with crewed aircraft.
IA systems could enhance existing UAS capabilities and enable new capabilities. For example, IA systems applied to UAS have the potential to do all of the following:
- Increase precision in agriculture by tailoring actions to specific situations, as in adjusting fertilizer or water applications within a particular field.
- Add flexibility to mission plans so that they better fit the situation and context, as in policing.
- Provide long-duration surveillance to better monitor critical areas, as in tracking chemical spills through a watershed.
- Take on dangerous missions currently executed by crewed aircraft, as in fighting forest fires.
- Facilitate coordination between different groups responsible for different aspects of a situation, as in urban firefighting teamwork.
- Learn from experience to adapt to patterns and preferences, as in fixed-obstacle avoidance during unmanned rotorcraft urban transit.
- Respond rapidly to sudden critical events, as in collision avoidance.
- Investigate from a distance, allowing experts to view local conditions, as in structural engineering inspections of damaged buildings following an earthquake.
- Act at a distance, allowing people to stand off from hazardous conditions and yet act to assess and respond to needs, as in response to industrial disasters with chemical, biological, structural, or radiological risks.
- Investigate from within, as in building interior inspections.
- Manage communication assets to maximize reliable connectivity and bandwidth utilization, as in replacing cell towers damaged in natural disasters.
- Make it easier and more affordable to observe and inspect large areas, as in the inspection of long-distance pipelines.
- Improve safety, as in general aviation, including light sport aircraft and amateur-built aircraft.
- Enable operations beyond the line of sight (that is, without continuous communication with ground personnel), as in monitoring the environment in remote locations.
5 D. Learmount, 2013/2014, When will we drop the pilot?, Flight International, December-January.
- Increase safety in off-nominal flight conditions by exploiting precision and rapid response characteristics to expand an aircraft’s operational envelope, as in rotorcraft operating in congested airspace.
The design space for unmanned aircraft is much larger than that for crewed aircraft. The absence of human beings removes many limitations associated with human frailties, from physiological limitations (e.g., acceleration limits, motion sickness, and other disorientation susceptibilities) to environmental adaptivity (e.g., pressurization and oxygen content) and more. It opens up a number of aircraft design options that can expand the operational envelope in a number of dimensions.6
- Long endurance. This capability has already been realized in a number of current operational systems that are primarily used for surveillance. The ability of a platform to stay aloft for periods that far exceed typical crew endurance has been demonstrated in a number of unmanned aircraft.
- Size/scale. This trend is also already under way, exploiting aircraft that are physically diminutive in comparison to even the smallest human pilot. Small unmanned aircraft continue to gain prominence and increased presence in combat applications and are being considered for civil and commercial applications ranging from law enforcement to pizza delivery.
- Maneuvering performance and agility. Unmanned platforms are inherently capable of sustaining accelerations well beyond the tolerance of the most fit and adaptable human pilots, improving maneuverability to unprecedented levels.
- Unique configuration options. There is no need to accommodate the constraints on size, volume, or shape imposed by an onboard human presence, increasing the possibilities for innovative airframe shapes and configurations.
- In-flight orientation. Vehicle orientation in flight can be completely arbitrary at any time during the flight profile, limited only by mission needs and considerations. Novel exploitation of this attribute, coupled with the increased configuration options mentioned above, can lead to innovative concepts that capture the best of fixed- and rotary-wing designs in a single configuration.
- Design lifetime. Small unmanned aircraft can be so inexpensive that they become economical even if they are designed to have a relatively short lifetime. A short lifetime is untenable for crewed aircraft, but it becomes a practical alternative for a small unmanned vehicle that could serve in a number of new applications and remake the conventional life-cycle cost paradigm.
- Novel launch and recovery methods. With IA systems, options for unconventional launch and recovery methods increase dramatically. These concepts would add new options for achieving runway independence and reduce infrastructure costs for a range of mission applications.
Current ATM procedures vary little from those followed by air traffic controllers 50 years ago. While automation has increased the availability of information and facilitates communications among controllers, decisions are still made by human beings. Some automatic alerting functions, such as conflict alerts, are designed to warn controllers when situations arise that need their attention, and some decision support tools assist controllers in determining the suitability of potential courses of action. Using the Observe, Orient, Decide, and Act (OODA) loop as a template, IA systems might be used in air traffic management as follows:
- Observe. Scan the environment and gather data from it. IA systems could assist controllers by monitoring many more data sources than humans can effectively monitor.
6 M.S. Francis, 2011, Unmanned air systems—Challenge and opportunity, Journal of Aircraft 49(6): 1652-1665.
- Orient. Synthesize the data into information. IA systems could monitor voice and data communications for inconsistencies and mistakes; monitor aircraft tracks for deviations from clearances; identify flight path conflicts; monitor weather for potential hazards as well as potential degradations in capacity; and detect imbalances between airspace demand and capacity.
- Decide. Identify options and determine a course of action. IA systems could identify and evaluate traffic management options for controllers and flow managers and recommend a particular course of action.
- Act. Follow through on decisions. IA systems could use voice and/or datalink communications to relay controller decisions to pilots, traffic flow managers, and other controllers. This could be particularly beneficial given that controllers spend so much of their time communicating.
The effectiveness of IA systems for ATM functions would be enhanced by the inclusion of compatible airborne IA systems that facilitate the exchange of information between pilots and aircraft and controllers and ATM systems.
In addition, IA systems could be used as an interactive training tool for initial and recurrent training of air traffic controllers. This could reduce the cost and duration of controller training and increase its effectiveness.
Federal Aviation Regulations (that is, Title 14 of the Code of Federal Regulations) direct pilots to remain “well clear” of other aircraft that have the right of way. Self-separation of aircraft is achieved through each pilot’s responsibility to see and then avoid conflicting traffic by maneuvering so as to remain well clear. Unmanned aircraft will require detect-and-avoid capabilities to meet self-separation requirements. In addition, regulatory changes will be needed to permit IA detect-and-avoid systems to satisfy the intent of current see-and-avoid regulatory requirements. However, the use of detect-and-avoid systems need not be limited to UAS. Such systems could also be incorporated into crewed aircraft to enhance their ability to self-separate.
Well-trained and qualified pilots are a source of resilience that contributes to the ultrahigh safety and the efficiency of the commercial transport operations. Pilots provide resilience to mitigate the safety and operational risks posed by weather, traffic congestion, workload, human errors, and system malfunctions. These risks include the following:
- Changes in operational circumstances, such as revised air traffic clearances due to weather and traffic congestion.
- Errors made by pilots, dispatchers, maintainers, air traffic personnel, or equipment designers.
- Equipment limitations and lack of availability.
- Equipment malfunctions.
Appropriately designed IA systems with carefully considered human interfaces could mitigate each of these risks.
Rotary-wing aircraft are extensively used throughout the world in a variety of civil aviation roles, many of which cannot be performed by fixed-wing aircraft. Rotary-wing aircraft missions include short-haul commercial passenger and cargo transportation, disaster relief, search and rescue, medical evacuation, construction, logging, fire fighting, pipeline management and inspection, drug interdiction, border control, traffic management, law enforcement, and agricultural services.7 Rotorcraft are particularly important for operations to destinations such as oil platforms, hospitals, and accident scenes without runways for fixed-wing aircraft.
The role of the rotorcraft pilot has been continuously influenced by the instantiation of IA systems. These systems have significantly reduced pilot workload and increased operational effectiveness and safety. Systems of particular interest include automatic turbine engine/rotor speed governing systems, stability augmentation sys-
7 Sikorsky Archives News, January 2014, Igor I. Sikorsky Historical Archives, Inc., Stratford, Conn., http://www.sikorskyarchives.com/pdf/news%202014/News%20Jan2014sm.pdf.
tems, automatic flight control systems, velocity and altitude hold systems, station-keeping systems, and autopilot and autonavigation systems. The accelerated use of IA systems can be expected to greatly facilitate the role of rotorcraft in the coming decades. In the process, the role of the human operator may be transformed from pilot to aircraft manager.
Optionally piloted rotorcraft are already under development for civil and military applications. Such aircraft could allow pilots to act as medical technicians or rescue crew in an emergency and to manage both their own aircraft and companion UAS. Optionally piloted rotorcraft would also reduce crew size, which would increase payload capacity and lower crew and training costs.
General aviation operations encompass a wide variety of equipage, pilots, and missions. Many older general aviation aircraft are equipped only for operations under visual flight rules (VFR), while many newer aircraft have redundant autopilot and flight management systems with glass cockpit displays. Pilots range in skill level from student pilots to commercial pilots, flight instructors, and air show performers with thousands of flight hours. As already noted, the accident rate for general aviation is much higher than that for commercial air carriers. Over the past decade, about 70 percent of the fixed-wing general aviation accidents were attributed to pilot-related causes, mostly associated with flight planning and decision making, particularly in high-workload situations associated with takeoff, approach, low-altitude maneuvering, and adverse weather.
IA systems have the potential to appreciably improve general aviation safety. Pilot-assist systems can facilitate in-flight decision making. Advanced autopilots can reduce the potential for loss of control through a combination of warnings and recovery options analogous to stability augmentation systems found in cars. Depending upon the design of the system and the nature of the emergency, IA systems could be designed to help a pilot avoid or recover from a loss-of-control situation either with or without direct input from the pilot. IA systems with navigation, detect-and-avoid, and flight planning and guidance capabilities would have the potential to eliminate the majority of general aviation accidents occurring today.
IA systems for general aviation would need to be inexpensive to support widespread adoption. However, apps currently available for tablets and smartphones offer better situational awareness than legacy general aviation avionics. Such tools are increasingly migrating into the cockpit as real-time situational awareness and decision aids. Autonomy research for civil aviation can improve the capabilities and quality of such tools as well as the process by which they are tested and certified.
UAS that currently exist or that could easily be developed can execute a wide variety of missions. These missions make use of a wide range of unmanned aircraft types, from diminutive micro air vehicles to large, high-speed aircraft the size of commercial transports. Some missions would be executed by fixed-wing unmanned aircraft, while others would be best served by unmanned rotorcraft with one, two, four, six, or eight rotors8 or by airships of various sizes.
Small, low-cost unmanned aircraft could inexpensively execute a wide variety of local observation and inspection missions, such as traffic and crop monitoring. Larger unmanned aircraft could execute missions with requirements for long endurance and/or large area coverage, such as inspection of oil or natural gas pipelines, wildlife observation, or communications relay. Large unmanned aircraft could also execute missions involving heavy payloads, such as advertising, crop dusting, construction, and fire fighting. Unmanned aircraft of various sizes could be flown by the general public for general aviation and personal recreation and by industry, academia, and the government for research, development, test, and certification.
The United States lags behind many other countries, such as Brazil, Uruguay, Chile, Japan, South Korea, Israel, and the United Kingdom in the commercial use of UAS, particularly in agriculture.9 This is noteworthy
8 Rotorcraft with four, six, and eight rotors are commonly referred to as quadcopters, hexcopters, and octocopters, respectively.
9 Y. Kim, “Grow More and Make It Affordable,” presentation to the committee on November 13, 2013.
because agriculture is perhaps the largest potential market for UAS and is one major global economic activity where they can markedly lower costs and increase productivity. Led by Japan and Brazil, unmanned aircraft are used to precision-spray pesticides and fertilizers; more than 90 percent of crop dusting in Japan is done by unmanned aircraft such as the RMAX unmanned helicopter, which is remotely piloted without the aid of any autonomous systems. Japan’s Administration of Agriculture oversees UAS operations there and requires annual and quarterly recurrent training of RMAX operators. Small unmanned aircraft equipped with cameras and other sensors can play another important role in agriculture by surveying crops to monitor growth, disease, and the application of pesticides, fertilizers, and water. 10 As a result, resources are applied where they are most needed, leading to production that is more economical and easier on the environment. Key technologies such as geofencing11 enable aircraft to stay within the confines of a specified region, including altitude limits. Many of these algorithms are being developed in open source and are readily available on the Internet.12
The ability of UAS to operate in harsh, remote environments with no risk to human life facilitates environmental research; a remotely piloted Global Hawk unmanned aircraft was recently used in Canada for ground mapping the Arctic and studying weather phenomena.13 Unmanned aircraft have also been used to monitor Japan’s Fukushima nuclear power plant accident in places too dangerous for humans.
The entertainment industry is also capitalizing on the use of UAS; when equipped with high-quality cameras, UAS are a viable alternative to expensive and sometimes dangerous filming from crewed helicopters. In March 2013, Paramount Pictures used a swarm of unmanned aircraft to create a Star Trek emblem over the Tower Bridge in London to advertise an upcoming movie. Attention has recently focused on the use of unmanned aircraft for high-end real estate marketing by capturing aerial views of estates both inside and out. However, the FAA prohibits the commercial use of unmanned aircraft in the United States without proper certificates of waiver or authorization,14,15 and it has shut down some innovative uses of UAS, such as on-site delivery of beer to ice fisherman by a brewery in Wisconsin.16 In addition to these regulatory impediments, low-altitude operations in cluttered airspace present a very difficult operational challenge as well as a correspondingly large commercial opportunity for the use of small UAS.17
Aviation is by no means the only domain to exploit increased autonomous capability. The use of increased autonomy in several other application domains has been significant and will continue to influence the development of IA systems with relevance to civil aviation. Nonaviation applications of IA systems have been largely driven by a few key factors:
- Missions, functions, or tasks that cannot be performed by a human (e.g., real-time control of spacecraft during entry, descent, and landing on Mars).
10 Associated Press, 2013, Agriculture the most promising market for drones http://www.foxnews.com/us/2013/12/14/agriculture-most-promising-market-for-drones.
11 A geofence uses GPS to check that an unmanned aircraft is within its designated operating area. If the aircraft approaches or exits this area, a return-to-base instruction is automatically executed to bring the aircraft back inside its designated operating area, which is defined by minimum and maximum altitude as well as by lateral (latitude and longitude) constraints.
13 Northrop Grumman, NASA fly Global Hawk in Canadian airspace for first time to study Canadian Arctic, PRNewswire.com, December 19, 2013, http://www.prnewswire.com/news-releases/northrop-grumman-nasa-fly-global-hawk-in-canadian-airspace-for-first-timeto-study-canadian-artic-236562771.html.
14 New York Times, Still unconvinced, home buyer? Check out the view from the drone, December 23, 2013, N.Y. Region, http://www.nytimes.com/2013/12/24/nyregion/still-unconvinced-home-buyer-check-out-the-view-from-the-drone.html?_r=0.
15 FAA, 1981, Advisory Circular 91-57, Model Aircraft Operating Standards, June 9.
16 Liz Fields, 2014, FAA slaps down drone beer delivery service to ice fishermen, ABC News, January 31, http://abcnews.go.com/US/faaslaps-drone-beer-delivery-service-ice-fishermen/story?id=22314625.
17 NASA, 2014, Enabling Civilian Low-Altitude Airspace and Unmanned Aerial System (UAS) Operations, Unmanned Aerial System Traffic Management (UTM) Workshop, NASA Ames, February 12-13.
- Nature of the mission or task (e.g., monitoring tasks that are so dull that it is difficult for humans to maintain a high state of alertness).
- Impacts and/or risks associated with relying on humans instead of IA systems.
- Economic and safety benefits.
Today’s fielded autonomous systems are largely made up of subsystems capable of performing individual functions or tasks autonomously and are often not integrated into accomplishing more complex tasks. As a result, overall mission capabilities are often realized only with considerable human oversight. As in aviation, future systems are likely to be far more functionally integrated, requiring only high-level supervision on the part of human operators. Over time, these systems will likely focus on accomplishing more complex tasks in more uncertain environments and be capable of effective contingency management for a wide range of anomalous conditions. As a result, they will require less and less continuous human cognizance and control.
The largest arena for the use of IA systems has been ground-based applications, covering a wide variety of stationary and mobile systems. Industrial manufacturing robots, consisting largely of stationary machines, were among the first to employ high levels of automation, having entered the marketplace in numbers 30 to 40 years ago. They have progressed from relatively simple, single-function machines (e.g., programmable milling machines) that required a well-defined operating environment to highly sophisticated multifunction machines capable of complex assembly and highly precise fabrication tasks under conditions of considerable uncertainty in geometry or environment. However, as sophisticated as these robotic manufacturing machines may be, they do not possess—nor do they need to possess—the sophisticated analytical capabilities that will be characteristic of the advanced IA systems for civil aviation.
Autonomy in the medical arena has been mostly evident in automated patient monitoring systems of various types. Although touted widely, medical robots today are mostly teleoperated systems guided by trained, skilled medical professionals who can add a high level of precision in certain surgical procedures.
Mobile systems (mobile robots, automobiles) that employ IA capabilities afford the greatest opportunity for synergy with the aviation world. Terrestrial mobile robots are used for diverse missions ranging from mine clearing and ordnance disposal to distribution of food carts in hospitals. The more sophisticated systems can perform complex physical tasks (obstacle avoidance and/or traversal, navigation in unknown environments, tactile object grasping, and autonomous inspection), while the simpler are typically focused on a single task and use relatively primitive algorithms. Most of these systems today involve a moderate to high level of human user monitoring or interaction. Robots for ordnance clearing and disposal employ sophisticated control and navigation systems traversing geographically challenging (uneven and uncertain) terrain, but they are largely teleoperated to accomplish their missions.
Automobiles have experienced the most remarkable growth in increased autonomy over the last decade. Since GM introduced the first electronic control unit in 1977, automobile control and monitoring systems have been transformed from almost purely mechanical units to the complex networks of microprocessors they are today. A typical luxury car now requires 70 to 100 electronic control units and over 100 million lines of computer code. This contrasts with 6.5 million lines of code for the avionics and onboard support systems on the Boeing 787 airplane.18 Software and electronics may account for 40 percent of a new car’s cost and 50 percent of warranty claims. This computer infrastructure that underlies modern vehicles enables a rapidly growing set of functions that includes IA cars.
These computers are already changing what it means to drive. They enable cars to take over many important driving operations, with features such as automatic parking and autonomous braking. Add to that lane-keeping assist and speed regulation systems, and the automation might exert more control of the vehicle than the driver. Already Volvo’s City Safety automatic braking system has reduced injury-related crashes by 18 to 33 percent
according to one estimate, 19 and the 2014 Mercedes S-class can drive at highway speed without input from drivers for short periods. The computer architectures in most production automobiles today feature federated systems, each of which specializes in the performance of individual automated functions, each one targeting specific attributes, such as fuel economy (e.g., fuel regulation) or safety (e.g., antilock brakes). In the last few years, mission-level autonomy has emerged to offer even more advanced capabilities, one of which is adaptive cruise control, which maintains vehicle-to-vehicle spacing.
The Google car initiative is further advancing and integrating these capabilities to provide a driverless, autonomous vehicle. Nissan has committed to the production of commercially available “autonomous” vehicles by the year 2020, but these vehicles will still require drivers to take action in some circumstances, and the full set of driver roles and responsibilities remains to be determined.20 In any case, in the future a driver’s primary task might not be driving but rather monitoring vehicle automation and interacting with information and entertainment systems. This shift introduces new vulnerabilities that can compromise driving safety.
As the technologies continue to mature with field testing and exposure to the driving public, they can be expected to gain increasing acceptance in a competitive marketplace that values economics, safety, novelty, and even prestige. Automobiles are not the only benefactors of IA capabilities: Trucks that operate without a driver could have obvious commercial value. But for many people, the attraction of autonomous systems lies in the chance for drivers to become passengers who can attend to work or enjoy new onboard entertainment and social networking systems. The safety benefits associated with some collision warning and collision mitigation systems, such as autonomous braking and stability control, also inspire the hope that vehicle-related fatalities can be greatly reduced.21 As with civil aviation, the use of IA systems in ground vehicles is supported by rapidly improving sensor and computer technology.
The promise of IA cars faces several important challenges. These include complex hardware and software failures that traditional verification and validation (V&V) processes might not address, cybersecurity vulnerabilities that threaten vehicle control, electromagnetic interference that could corrupt signals, and human–machine integration issues.22 Of these, the latter represent a particular challenge because one of the main appeals of IA vehicles is that they will relieve the driver of the need to continuously attend to the road and will enable him or her to engage in other activities. Although vehicle automation systems may accommodate the least demanding driving situations and encourage drivers to disengage from the driving task, they will also demand that a driver be able to reengage in the control loop, sometimes more quickly than is humanly possible. Drivers are most likely to fully rely on automation during long-duration highway driving, when a driver will need to very rapidly reenter the control loop if something suddenly goes wrong. This poses a safety challenge, a legal challenge regarding presumed responsibility, and a social challenge with respect to acceptance of the new technology.
Although ground vehicle applications of autonomy for the most part differ substantially from civil aviation applications, there are also several parallels and opportunities to build joint expertise on responding to the challenges of increased autonomy:
- The drivers of cars and the pilots of general aviation aircraft would both benefit from relatively inexpensive IA systems that could to some degree serve as copilots, alerting the driver or pilot of hazards that have been overlooked and taking corrective action in extremis.
- V&V of automobiles with IA systems face challenges similar to those faced by civil aviation. The National Highway Traffic Safety Administration (NHTSA) does not regulate or certify vehicle designs in the way that FAA regulates and certifies aircraft designs—and FAA certification standards tend to be much more
19 Volvo, City Safety loss experience—An update, Highway Loss Data Institute Bulletin 29(23): 1.
20 Nissan boss wants autonomous cars by 2020, Autocar, March 7, 2014, http://www.autocar.co.uk/car-news/geneva-motor-show/nissanboss-wants-autonomous-cars-2020.
21 Volvo, City Safety loss experience—An update, Highway Loss Data Institute Bulletin 29(23): 1.
22 National Research Council, 2012, The Safety Challenge and Promise of Automotive Electronics: Insights from Unintended Acceleration, Transportation Research Board Special Report 308, Washington, D.C.: The National Academies Press.
rigorous than the safety standards that must be met before IA systems can be introduced into cars.23 Even so, NHTSA and the FAA may benefit from sharing information regarding their efforts to assure that IA systems are safe and reliable. In particular, the FAA may benefit from operational data collected on the performance of advanced IA systems deployed in cars.
- Because both aircraft and ground vehicles operate in challenging physical, electromagnetic, and social environments, they may face similar vulnerabilities. These vulnerabilities range from unusual electrical failures (e.g., tin whisker shorts) to cyberphysical security.24
- Issues related to coordination of control in ground vehicles may be relevant to aircraft. The safety and performance of automobiles depend on the surrounding vehicles. The network effects of vehicle automation that propagate across a traffic stream could dominate vehicle response in some situations. This may become an even stronger influence with the connected vehicle concept, which provides data links among different vehicles and between vehicles and the infrastructure. Civil aviation may face a similar situation as aircraft communicate and interact more directly with one another.25
Unmanned surface maritime vehicles are relatively new to the robotic family. Most contemporary systems are highly operator intensive (i.e., teleoperated), but they do exploit IA technologies such as augmented stabilization and control in high seas and during waypoint navigation. Advanced unmanned underwater vehicles are acquiring higher levels of autonomous functionality to enable extended periods of unattended operation. Employing autonomous navigation and other autonomous robotic functions, the relatively slow, deliberate movements and transit speeds of unmanned underwater vehicles help ensure that they have ample time to process information for autonomous decision making. Unmanned surface vehicles, which may operate at much higher speeds than underwater vehicles, also need to detect and avoid other watercraft. Marine vehicles operate in an environment with no inherent structure, such as roads.
Space missions traditionally rely heavily on humans for mission-level decisions, even in the case of robotic missions with sophisticated satellites or planetary exploration rovers. To reduce manpower requirements and account for the time delays in communications, the International Space Station (ISS) incorporates advanced smart sensors for failure recognition, diagnostics, and prognostics; model-based reasoning for scheduling maintenance; and automation of low-level routine tasks (e.g., directing imaging sensors). However, many advanced in-space functions such as inspection, maintenance, and assembly on board the ISS still require teleoperation or supervision by a human operator. IA technologies for deep space missions are required for tasks such as autonomous precision landing, and they greatly facilitate the operation of planetary rovers, enabling them to localize and avoid hazards and to execute functions that cannot be efficiently controlled from the ground because of communication delays. The Mars’ Curiosity rover operates with scripted plans provided by ground personnel. The rover executes the plan using sensors to detect and control the vehicle to navigate terrain and avoid obstacles. Advanced technologies in space systems include real-time and life-cycle vehicle health information to enable advanced decision-making algorithms to inform logistics management and maintenance planning.
23 NHTSA provides guidance to manufacturers through the Federal Motor Vehicle Safety Standards, which set minimum performance requirements but do not specify how manufacturers should meet those requirements. NHTSA assesses compliance by inspecting and testing sample vehicles. It also monitors for safety defects by examining consumer reports of anomalous vehicle behavior that might pose a safety risk.
24 NRC, 2012, The Safety Promise and Challenge of Automotive Electronics: Insights from Unintended Acceleration, Transportation Research Board, Washington, D.C.: The National Academies Press.
25 NHTSA is taking steps to enable vehicle-to-vehicle communication technology for light vehicles. It has recognized the potential of this technology to improve safety by allowing vehicles to communicate with each other and ultimately avoid many crashes altogether by continuously exchanging basic safety data, such as speed and position.
Space applications of autonomy may contribute new algorithms for a variety of functions that would benefit the civil aviation community. These include route planning, autonomous navigation, obstacle avoidance, autonomous landing site selection, and automation of lower-level maintenance and operational functions. Another contribution will come in the form of the crewing concepts needed to manage a constellation of vehicles versus the single-platform model now widely used. Space systems already employ a different crewing paradigm than typical crewed systems. Perhaps the most significant application of autonomy to civil aviation will come from the technologies used to deal with the long time delays between the ground and the spacecraft. The space domain has learned to use robust software algorithms to successfully operate spacecraft without continuous human inputs.