Effects of Autonomous Vehicles on law enforcement
Mass-adoption of self-driving cars will completely reshape the nature of police interactions, from ticketing to remote control
Introduction
Large-scale deployment of AVs on our roads will certainly have an impact on multiple stakeholders: from drivers or pedestrians to OEMs. A key, yet often overlooked actor, are law enforcement authorities.
As the deployment of AVs helps increase safety and compliance on the roads, it is also likely to have a tremendous impact on police forces’ size[1]. AV’s will also reshape the nature of the interactions concerning police authorities. From ticketing or towing to remote-control, the objective of this memo is to explore how mass-adoption of self-driving cars might affect these interactions[2].
Ticketing
In March 2018, a San Francisco police officer issued the first traffic ticket to an AV. The car, a GM’s Cruise, was on self-driving mode when it failed to yield a pedestrian at a crosswalk.
A traditional approach (as it happened with the Cruise) is to ticket the operator behind-the-wheel. However, in the future, it is likely that drivers will abandon AVs in favor of passengers. Who would be ticketed then? More interestingly, what would happen if nobody is in the vehicle?
Conversations with experts, seem to point in the same direction. Similarly to how it is done in ATC[3], a wide array of agents could be subject of fines:
- Operators/passengers: Operators could face fines to the extent the car is not being operated in fully autonomous mode. For instance, the operator of a SAI L4/L5 car could be to blame if the systems had been inappropriately “overridden” (speeding due to fabricated medical emergency)
- Fleet owners: As a handful of players concentrate AV ownership, fines could be issued both on a case-by-case basis and by analyzing the overall incidence of violations (waiting time in load-only zones)
- AV companies: Generalized malfunctioning or failure to adapt to local regulations could also trigger sanctions for developers
In any case, as AVs become widespread, it is unlikely that police officers will focus their efforts on catching minor violations by driverless cars. Instead, police will target the declining number of manually-operated cars. Furthermore, the vast amounts of data onboard AVs will discourage officers: AV vehicles will offer an unprecedented level of objectivity.
Lastly, all experts agree on the inevitable decrease of ticketing and traffic-related revenue. For instance, NYC alone collects a yearly average of $1.8 Billion in fines and fees[4], with most coming from driving-related violations[5]. The significant diminish of these revenue streams will certainly cause fiscal imbalances in cities like San Francisco ($512[6] in fines per capita), D.C. ($502) or Chicago (>$248). Whether municipalities will be able to introduce urban roads usage, or congestion fees to partly offset this effect, remains to be seen.
Pullovers and interactions with AVs
To address this issue, Waymo released in May 2018 the “Law Enforcement Interaction Protocol”: a 36 page handout with instructions on how police could immobilize its vehicles in emergency situations. Through cameras and audio sensors, Waymo’s cars are designed to pullover if they detect a police or emergency vehicle flashing from up to 100 feet away. After stopping, the car is unlocked and the windows roll down automatically, allowing Waymo’s support team to interact with law enforcement representatives via an intercom system.
In the handout, Waymo also provides agents with alternatives on how to immobilize the car (from breaking one window to cutting the battery cables) and methods to tow the vehicle or safely extract passengers. More importantly, it provides a 24-hour hotline number, so that the car can be remotely controlled and shutdown.
Unfortunately, most AV companies lag behind Waymo. Last December, Redwood City police patrol intercepted a sleeping Tesla driver on the highway, after repeated flashes and attempts to wake up the driver in vain, police improvised a strategy by driving in front of the Tesla and performing a brake-test. The Tesla emergency braking system engaged and the vehicle stopped promptly.
Looking forward, and, as manually-driven police patrols have to cope with an increasing mass of AV cars, it is foreseeable that Waymo’s model will prevail: one in which cars will automatically stopped if flashed, and could be remotely turned off from an emergency hotline. As the AV industry becomes more fragmented and more AVs are on the roads, experts point out to the potential development of three-way V2X protocols. Regulators have already explored this space with precedents such as EU’s eCall[7].
Remote control
In March 2018, a draft by federal DOT, industry groups and the RAND was revealed to the press, its main thesis pivoted around the idea of preventing potential massive cyberattacks targeting AVs. Experts who participated on the panel, suggested the possibility that authorities could interact and even control these vehicles.
Although far from these demands, Waymo’s recent attempt to provide authorities with guidance on how to immobilize its vehicles, has been received with good eyes.
Similarly, remote controlling of AV cars could help fight car thefts. A more futuristic and interventional approach to remote controlling is being explored by the Dutch Police[8]. Imagine for a second that you have your AV stolen. After notifying the police, authorities request permission to “takeover” the vehicle. Shortly after hacking into your AV systems, the car drives itself into custody to the nearest police station, with the thief locked inside.
As opposed to pullover systems, remote control seems less likely to appear in the short term. The main concern is not only privacy: experts argue that these tools, originally built around public safety, could be exploited by terrorists to cause even more harm.
Forensics (accident investigations)
Modern AVs such as Tesla’s Model S feature a device called EDR/AD[9], an analogous car-version of a plane’s Flight-Data-Recorders. These automotive “black-boxes” have been equipped on commercial vehicles for decades. In AVs, however, the technology onboard has enabled a much more accurate recreation of the exact situation right after an accident.
Data collected by EDRs will become instrumental in assessing who is responsible for a crash, offering an unprecedented level of objectivity to police authorities, insurance companies and citizens. An expert from Mountain View Police Department described seeing a “reconstruct” on a Waymo car as “heaven”. In his words, the recorded data ─which can be visualized onboard or outside the vehicle─ felt like “watching instant replay on the NFL”.
Tesla’s EDR data (see below) on the driver’s attention to the car systems ─the driver barely touched the steering wheel in the 37 minutes preceding the crash─ was key in leading the NHTSA to conclude that the 2015 crash was an operator mistake rather than an Autopilot failure.
To the extent that AVs become safer and its accidents extremely infrequent, it is also likely that these accidents could be investigated by a national board (probably NHTSA) on a case-by-base basis in a similar fashion to how NTSB investigates plane crashes.
Most concerns around EDR/ADs data are around data-extraction mechanisms and the legislation surrounding them. In California, CalECPA[10] laws require a warrant be issued for police to extract data AV. Although California’s DMV is reportedly working on a standard, it will be interesting to see how these policies develop, and if they will be escalated to the federal level.
Autonomous Surveillance
In previous topics, we have discussed mostly reactive approaches to how police could react to the deployment of AVs. However, throughout different interviews, some experts raised the possibility that law enforcement authorities could take a more audacious approach.
One of them, is the possibility of leveraging AVs to conduct autonomous surveillance and even arrests. To enlighten this point, Motorola recently filed a patent for autonomous police cars. The patent sketches a driverless police car with built-in fingerprint sensors and a breathalyzer that could identify suspects on the street. If the suspect obeys, and it results on a positive identification that requires a hearing or trial, the AV reads the suspect’s rights and allows it to contact a lawyer and a judge via videoconference. The AV would offer live translation in the detainee’s mother tongue and include a payment processing device in case the violation results in a fine or bail.
Conclusion
- Overall, AVs are likely to change the size of police forces in the long-run: some experts argue that, with mass-adoption of AVs, police workforce could be cut in half without a major compromise in safety
- In the immediate future, AV companies will follow Waymo’s lead, and develop police interaction protocols, including automatic pullovers and remote shutdown
- Ticketing will be extremely rare for AVs, and could generate fiscal imbalances in multiple cities — if they are not offset by usage and pollution fees —
- EDR/ADs will transform automobile forensics, offering an unprecedented level of objectivity and detail to multiple stakeholders
- Remote control of AVs and autonomous policing is unlikely to appear in the medium-term due to technological, privacy and security concerns
Footnotes:
[1] Source: “Bureau of Justice Statistics (BOJ) surveys: Contacts Between Police and the Public, 2015”
[2] Note: This memo does not deal with issues such as impaired driving or potential terrorist cyberattacks, which are covered in other memos
[3] Air Traffic Control
[4] 2012–2015
[5] Parking, driving, and pollution
[6] Source: “No One at the Wheel: Driverless Cars and the Road of the Future”, Samuel I. Schwartz
[7] EU mandates every new car to feature eCall free-of-charge, a system that automatically connects passengers with police and emergency services after a crash
[8] Extracts from interview with Chief Innovation Officer for the Dutch Police
[9] Event-data-recorder for autonomous driving
[10] California Electronic Communications Privacy Act (CalECPA) — SB 178
