By John Shields
In 2004, the Defense Advanced Research Program Agency (DARPA), with the backing of Congress, created the DARPA Grand Challenge. The object of the challenge: create an autonomous vehicle that could navigate a desert route between Barstow, California and Primm, Nevada.1 The first year, not a single entry succeeded. In 2005, the competition was held again; this time five teams succeeded and a team from Stanford won the $2 million prize.2 In 2007, a new course in Victorville, California was designed to test the ability of a vehicle to not only navigate urban obstacles, but to follow local laws while doing it. 3
Nearly 15 years later, autonomy is not fully realized. To be certain, the goal of automating one-third of the ground combat vehicles set out by Congress has not been realized.4 Despite this shortfall, DARPA started a revolution in autonomy that has companies such as Google5, Tesla6, Uber7, and Mercedes8 working on the problem of autonomous vehicles. Not far over the horizon is a day when autonomous vehicles (passenger vehicles, busses, trains9, semi trucks10) are the norm.
When DARPA sped up this revolution, the full implications were probably not considered. The current environment of terrorism and rogue attacks raise serious questions about the defense of installations around the globe. Autonomous vehicles will change the threat environment in numerous ways. What does that threat environment look like now and how do active and passive countermeasures change when autonomous vehicles are introduced as the primary mode of transportation?
An emphasis on the nefarious uses of autonomous vehicles by hackers has misled the public into believing that if we can only secure the infrastructure and software, the physical security threat will be nullified. This flawed way of thinking has led people to focus on nefarious uses of autonomous vehicles that require technological know-how. The true threat is that autonomous vehicles, when fully adapted, will provide a simple yet effective delivery system for an explosive device or biological weapon that requires no technological skills.
Contemporary deterrence
At military installations around the world access is granted by presenting a controlled form of identification which is reviewed for obvious tampering or scanned for a two-part authentication. For those who do not possess a qualifying identification, acceptable forms of identification (i.e., a driver’s license or passport) is reviewed and compared to an access roster. The only way to get on the access roster is to be sponsored by someone with full access to the installation. The system is not flawless, but it allows for freedom of movement and a discrete manner by which to verify individuals. On installations the world over, vehicles are required to be registered to a person who has access to a base. Combined, the two practices serve as a deterrence for would-be perpetrators. But this method is labor intensive, requiring armed military personnel, government employees, and contractors to man gates 24 hours a day.
Additionally, countermeasures such as active barriers, berms, fences, and other natural and man-made obstacles create physical deterrence that is difficult to overcome by even the most zealous of wrong-doers. Surveillance cameras and roving, random patrols on foot, in vehicles, and with military working dogs add depth to the defense of installations that further deters individuals and makes planning an attack difficult.
An emphasis on the next generation installation is moving forward across the DoD. At the vanguard of this movement are concepts such as the Marine Corps’ Installation–neXt.11 This program looks to reboot Marine Corps installations to reflect the modern nature of asymmetrical and irregular warfare. It aims to do so by integrating autonomy, artificial intelligence, analytics, and interconnected objects to increase efficiency and decrease wasted resources.
Over the horizon
As has been demonstrated by many companies, the future of autonomous vehicles is not to make humans better drivers, but to eliminate the error causing humans in the loop. It is estimated that every year nearly 35,000 people in the US die in automobile accidents.12 The National Highway Transportation Safety Administration (NHTSA) estimates that nearly 94% of all accidents are caused by human error.13 Industry has promised to reduce this human error through the implementation of autonomous vehicles.
Of course, it isn’t all altruistic. Companies stand to make a large profit by shifting their focus and providing safer, more productive uses of resources. Companies plan to do this through ride-hailing and car-sharing platforms, reducing vehicle ownership.14 Users will subscribe to a vehicle platform and be provided service based on availability and necessity, making car sharing the new norm.
Vehicle autonomy will not just enable car sharing, but will actually change the concept of a car. Companies such as Google have already toyed with the idea of removing the most common human-vehicle interfaces: the steering wheel and brake pedals.15 While this novelty in the tiny google golf cart-like vehicles made some uneasy, subsequent generations are born into autonomy and have no trouble adapting.
Elon Musk, the CEO of Tesla, has said that a car will be able to go from home to work without human input.16 Major automakers are following suit by incrementally decreasing the role of the human in passenger vehicles. Adaptive Cruise Control, emergency braking, distracted driver alerts, and blind spot warnings are just some of the features becoming ubiquitous on modern cars.17 And for good reason, the Center for Disease Control estimates medical costs associated with death or injury from motor vehicle crashes exceed $63 billion annually.18 People are lauding these efforts and many states are adopting policy to ensure these vehicles can be licensed and tested—the groundwork is being laid for a future without humans at the wheel.19
When the wolf wears sheep’s clothing
When reviewing the literature on threats of autonomous vehicles, the focus is split between ethics and malicious code.20 It is argued that in the hands of a hacker, a vehicle could be used as a getaway car, remotely driven through crowds, or as a way to take hostages. How an autonomous vehicle is utilized depends wholly on the imagination of the person with the skills to manipulate the vehicle’s software. But let’s imagine that an individual didn’t have to manipulate any code or use the vehicle in any way for which it wasn’t designed.
There is precedence for this type of use for autonomous vehicles in the form of drones. Off-the- shelf drones can deliver small, but effective, payloads to targets guided by GPS, or via remote control with very little to no modification. The disadvantages of drones are their inability to blend in because of their visibility, the noise they make, the small payload capability, and the relative ease with which they can be defeated with fairly simple countermeasures. Fast forward ten years and these facts are likely to stay the same for drones. In that type of environment, the wolf can’t blend in.
Autonomous vehicles will become an accepted part of society. Vehicle ownership will wane and among many of those who will partake in the adoption will be individuals who live and work on a military installation. A vehicle coming toward a gate without a driver will seem odd at first, but before long it will become the norm. When the first autonomous vehicles show up at military installations, there will be confusion on the part of the car and the guard. The guard is not trained to grant access to a vehicle without a driver. The vehicle does not have a Common Access Card that can be scanned for two-part authentication.
Just as guerilla fighters intentionally blend with the people they live amongst, autonomous vehicles will all appear the same. Mundane and mass produced, autonomous vehicles may be the ultimate guerilla fighter, they have no emotion, relatives to think of, or opposition to oust them before a crime is committed. This new norm allows the wolf to don the fleece.
The use of autonomous vehicles will not be limited to transporting people. Autonomous vehicles promise to be a major player in the personal package delivery market. Packages can be placed in an autonomous vehicle and delivered to a destination while enroute to drop off people. Because of this, it may be common to see a vehicle with extra packages. It would be impossible to have a clearinghouse for all the packages being delivered as the efficiency is gained in a person not having to go anywhere to send a package. Recently, the car manufacturer Ford released video showing an automated car that carried a delivery robot. The robot, with human-like upright mobility, can carry a package from vehicle to doorstep while navigating obstacles along the way.21
Unlike the fanciful ideas proposed for nefarious uses of autonomous vehicles, an autonomous vehicle loaded with explosives needs no martyr and need not drive erratic if it blends in. It must simply be directed to an address or a GPS coordinate. Because a driver is not necessary, human intervention is further removed making killing from a distance equivalent to playing a video game. Further, an individual could be paid to hail a car and load it with an explosive device without realizing.
Targets within an installation become a more tenable prospect because the ability for a car to arrive at a specific location at a specific time will undoubtedly be one of the benefits of autonomous vehicles. Currently, the gate of an installation is attacked because of the uncertainty of getting past it. Killing the unknown individuals at the gate will spark frustration, but penetrating the perimeter to hit a well-known target will cause confusion and chaos. The burden of detection is disproportionately borne by the installation, causing a level of scrutiny that, in and of itself, may be a force divisor. Inherently, there will be a disproportionately low level of risk for the perpetrator.
The subscription model of autonomous vehicles will make it easier for law enforcement to identify perpetrators through accounts, payment methods, and location data of the vehicle. However, if the lessons of cryptocurrency are any indication, anonymity will likely be an inherent precept of many future ride-hailing services. Being able to detect these types of incidents will require a significant number of policing resources in an already tight budgetary environment.
Undressing the wolf
The threat posed to society by autonomous vehicles is outweighed by orders of magnitude for how society will benefit. Autonomous vehicles are inevitable. Whether Tesla’s “soon” is reasonable or it takes 20 years, the ability for the Department of Defense to manage this emerging threat to security must start now. Below are a few areas that should be explored:
On-the-fly chemical and biological detectors
An informative proxy for processing the volume that will be required when entering and exiting military installations is that of US ports of entry. In 2018, the Port of Los Angeles processed nealy 9.5 million containers.22 At ports around the US, containers are scanned to determine whether nuclear material is present.23 The process is used with automation to ensure expedient transfer of shipping containers from the boat to trucks and trains.
This same concept can be applied to scanning the contents of autonomous vehicles. However, there is a significant difference between detecting decaying isotopes and analyzing the chemical or biological compounds of explosives or biological weapons. To start with, there are only 37 known radioactive elements that each possess a unique signature in the form of the isotope’s half-life.24 There are thousands of different compounds that make up explosives, many of which are used in everyday products such as ammonium nitrate in fertilizer.
To maintain efficiency in the processing of autonomous vehicles through a checkpoint, the ability to check for chemicals must be done at a distance and must be safe for human exposure. Many techniques can be used, but they all have limitations such as the need for direct contact, the inability to “see” through other materials, or close proximity to the explosive (being in the center of a vehicle may put it out of range).25 “There are thousands of explosive compounds and no single approach is capable of detecting every such compound.”26 These technologies, though already in differing stages of development, will need to be implemented along with any other installation upgrades.
The physical approach
The physical environment will change dramatically in a world where autonomous vehicles are the norm. The lanes carrying autonomous vehicles onto installations will appear much like current tolling lanes on highways. Incorporated in these lanes will be the ability to quarantine or hold an autonomous vehicle for further scrutiny. The difficulty of these lanes, at least initially, is that installations will have to incorporate autonomous vehicle lanes and traditional lanes. There are a number of installations which will not have the space to expand to encompass both.
A physical or virtual standoff will be required for vehicles that are not validated to enter the installation. A physical standoff space would look much like the cell phone waiting area at most large airports. Finding space for what amounts to an autonomous vehicle parking lot will strain the physical resources of smaller installations. A better option would be a virtual standoff whereby vehicles not verified will be turned away until verification can be made.
Autonomous vehicles may approach only when verified by a Blockchain27 mechanism and automatic registration that passes authority for a vehicle kill switch within a perimeter.28 Vehicles that fail to meet this protocol will be denied entry into the boundaries of an installation. A permission based protocol must be developed and adapted by industry for just such purposes. This will apply more broadly to other locations such as homes, hospitals, banks, and drive-through establishments. Based on limited research, this type of protocol for autonomous vehicles is not being discussed.
Protocols, coding, and laws
Determining how humans will interact with autonomous vehicles will be key. Current deterrence tactics are formed based on the traditional model of a human at the wheel. This model requires armed guards to shoot at a driver, who will either stop due to a life-preservation response or die because of a martyr mentality. The ability to stop a vehicle driven by a human is therefore determined by the ability to disable the human.
Autonomous vehicles are programed to take human life into consideration, however, there are a great number of factors considered in the programming of the vehicle.29 An understanding of the programming criteria will be necessary for proper interaction with an autonomous vehicle. A protocol to determine what steps are to be taken if the vehicle malfunctions and does not stop is necessary for proper policing and force protection aboard installations. These steps will be a requirement for the Department of Defense as they adopt the technologies for use in military vehicles.
Autonomous vehicle laws and policies will differ by location and government. Vehicle manufacturers will likely adopt the regulations of the largest restrictive market (sometimes referred to as the California Effect).30 This may provide some standardization and common protocol, but research into how this may affect the ability of the military (or Department of State missions around the world) to review and authorize vehicles to enter the controlled space beyond the gate must be addressed. Early input by government regulators can shape this environment favorably in the development phase rather than in a reactionary manner as much of the policy on technology is instituted.
Conclusion
The adoption of autonomous vehicles by societies around the world will happen, though the timeline is in question, the most conservative estimates are within decades. This technology promises to be a panacea for the 94% of human error induced deaths each year in car crashes in the United States.31 Full autonomy, like most technologies, will likely only be available on vehicles with higher than average price points at first. But the economies of scale will drive down prices. Aging components of the vehicle fleet will likely be sold to countries where their use and protocols may be changed or manipulated. This poses its own unique set of challenges that should be concurrently addressed.
It is imperative that the DoD gets ahead of the commercial market on this topic to ensure continued security at installations worldwide. DARPA may be a great venue through which to explore the technical aspects of this cumbersome problem. The private sector will need these protocols elsewhere in the autonomous driving ecosystem to allay concerns of privacy and the nuisance of unwanted autonomous traffic. Through joint efforts between government and industry researchers, an effective solution can be developed.
At Marine Corps Air Station Miramar, courses have been laid out to test autonomous vehicles aboard military installations and in military scenarios.32 Through partnerships with companies such as Uber and Lyft, the potential for autonomous vehicles will be realized within the DoD much sooner than many realize. Platooning of vehicles, the ability for several autonomous vehicles to form a road train, promises to revolutionize logistical networks and decrease fuel consumption.33 The ability to take people out of the driver’s seat and put them behind weapon systems provides greater offensive power for the DoD and reduces the potential for injury or death on military convoys. After all, a vehicle programmed not to stop when fired upon will complete its mission unless physically incapacitated.
The author of the article describing the Installation–neXt program uses the parable of the city of Troy during a ten year siege to prove his point.34 The city withstood bombardment, attacks, and attempts to knock down the gate. Academics have a great number of theories about the most famous story to come out of the siege of Troy: that of the Trojan Horse.35 Widely acknowledged as a myth-turned-metaphor, its lessons should not be lost on this generation of technology adopters. Inviting a vessel into the city without knowing what it contains may be detrimental to everyone in the city.
The DARPA Grand Challenge set the wheels for autonomous vehicles in motion in 2004. Out of that challenge autonomous vehicles have been built that can ascend Pikes Peak at breakneck speeds, park themselves, avoid collisions, and change lanes.36 It would be a shame that a technology seeded by forethought in the National Defense Act of 2001 cannot be realized because of an impasse at the gate.
Footnotes
1.The Grand Challenge https://www.darpa.mil/about-us/timeline/-grand-challenge-for-autonomous- vehicles
2. The DARPA Grand Challenge: Ten Years Later https://www.darpa.mil/news-events/2014-03-13
3.DARPA and Drone Cars: How the US Military Spawned Self-Driving Car Revolution https://www.livescience.com/44272-darpa-self-driving-car-revolution.html
4. National Defense Authorization Act for Fiscal Year 2001 (S. 2549, Sec. 217)
5. https://waymo.com
6. https://www.tesla.com/autopilot
7. https://www.uber.com/info/atg/
8. https://www.mercedes-benz.com/en/mercedes-benz/next/automation/
9. The Future is Driverless https://www.railprofessional.com/magazine/december-2017/the-future-is- driverless
10. Self-Driving Trucks Are Now Delivering Refrigerators https://www.wired.com/story/embark-self-driving- truck-deliveries/
11. https://mca-marines.org/wp-content/uploads/2019/02/Installation—neXt.pdf
12. CDC Motor Vehicle Statistics https://www.cdc.gov/winnablebattles/report/motor.html
13. https://www.nhtsa.gov/press-releases/usdot-releases-2016-fatal-traffic-crash-data
14. Your Next Car Could Be a Flexible Subscription Model https://www.forbes.com/sites/sarwantsingh/ 2018/07/30/your-next-car-could-be-a-flexible-subscription-model/
15. Google Ditched the Steering Wheel Because People Are Unreliable https://www.engadget.com/ 2015/03/17/why-google-ditched-the-steering-wheel/? guccounter=1&guce_referrer_us=aHR0cHM6Ly93d3cuZ29vZ2xlLmNvbS8&guce_referrer_cs=OLcBxlXu_ 2hkVKgDaq1_0Q
16. Musk Says Commute With “No Driver Input At All” Coming ‘Soon’ https://www.autoblog.com/ 2018/12/10/musk-says-commute-with-no-driver-input-coming-soon/
17. Advance Driver Assistance Systems https://www.lifewire.com/advanced-driver-assistance- systems-534859
18. Cost Data and Prevention Policies https://www.cdc.gov/motorvehiclesafety/costs/index.html
19. The State of Self-Driving Laws Across the US https://www.brookings.edu/blog/techtank/2018/05/01/the- state-of-self-driving-car-laws-across-the-u-s/
20. Hackers are the Real Obstacles for Self-Driving Vehicles https://www.technologyreview.com/s/608618/ hackers-are-the-real-obstacle-for-self-driving-vehicles/
21. https://www.businessinsider.com/ford-delivery-robot-digit-walks-on-two-legs-like-human-2019-5?IR=T
22. https://www.portoflosangeles.org/business/statistics/container-statistics
23. https://www.dhs.gov/blog/2017/04/06/new-radiation-detection-system-operational-major-us-port
24. https://www.thoughtco.com/list-of-radioactive-elements-608644, https://www.nde-ed.org/EducationResources/HighSchool/Radiography/halflife1.htm
25. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6120018/
26. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6120018/
27. What is Blockchain? https://www.ibm.com/blockchain/what-is-blockchain
28. It is important to note that while Blockchain is currently a very secure mechanism through which to verify transactions, it is not impermeable. Nearly $2 billion of cryptocurrency has been stolen over the past few years and the money lost cannot be retrieved because of the irreversible nature of a Blockchain. https://www.technologyreview.com/s/612974/once-hailed-as-unhackable-blockchains-are-now-getting- hacked/
29. The Social Dilemma of Autonomous Driving http://science.sciencemag.org/content/352/6293/1573
30. https://e360.yale.edu/features/californias_clean_car_rules_help_remake_us_auto_industry
31. https://www.nhtsa.gov/press-releases/usdot-releases-2016-fatal-traffic-crash-data
32. https://www.roboticresearch.com/news/modern-installation-mobility-autonomous-vehicle-proving-ground/
33. https://www.sciencedirect.com/science/article/pii/S0167642317301168
34. https://mca-marines.org/wp-content/uploads/2019/02/Installation—neXt.pdf
35. http://www.ox.ac.uk/news/arts-blog/did-trojan-horse-exist-classicist-tests-greek-myths
36. Audi’s Robotic Car Climbs Pikes Peak https://www.wired.com/2010/11/audis-robotic-car-climbs-pikes- peak/
Convergence: Policy, Science, and Technology 