December 2018

This fall we’ve spent a fair amount of time talking about how connected and automated vehicles (CAVs) will change the structure of our cities, from the curb, to public transit, and beyond. In my last post before the holidays, I want to take a look at how CAVs could change the way goods are transported and delivered within cities. While they probably won’t reach Santa-levels of delivery efficiency, CAVs may help make last-mile deliveries more efficient (and could help fill the current shortage of truck drivers in the US, but that’s a subject for another day).

CAVs are already being tested as delivery vehicles by companies like Domino’s and Kroger, while earlier this year Toyota announced delivery partnerships with Amazon and Pizza Hut, and Waymo’s CEO recently highlighted it as an area of opportunity.  This week the New York Times profiled Nuro, the start-up working with Kroger to test robotic delivery cars in Scottsdale, Ariz. Nuro’s vehicles are designed in-house, and look like “toasters-on-wheels.” Currently they followed everywhere they go by human safety drivers in conventionally driven “shadow car,” since the vehicles are still in testing. When the vehicle stops for a delivery, customers enter a PIN code into a small touch pad to open a compartment containing their order. The current charge for same-day delivery using the system is around $6. Ford has also flagged the delivery market as an area they’d like to explore, citing projections that, by 2026, the last-mile delivery market for CAVs will hit $130 billion.

But the roads are not the only path automated vehicles may soon tread in their mission to bring you your takeout order. A number of companies, including Postmates, are working on delivery robots that will cruse down the sidewalk and roll right up to your door. Last year I even personally witnessed Postmates’ bot rolling along the streets of Washington. As exciting as it would be to have R2-D2’s cousin deposit an order of egg rolls on your doorstep, the deployment of delivery bots raises an interesting question of how much space we’re willing to give up to automated devices. The sidewalk is a human dominated space, and, especially in cities, is already busy with foot traffic. Will people be willing to cede some of this space to a robot? Yet another question that city regulators and individual citizens will be forced to answer as automation makes greater inroads to our daily lives.

P.S. – Last week a delivery robot caught fire in Berkeley, leading some locals to build a memorial in its honor.

By Jesse Halfon


Last month, two California Highway Patrol (CHP) officers made news following an arrest for drunk driving. What made the arrest unusual was that the officers initially observed the driver asleep behind the wheel while the car, a Tesla Model S, drove 70 mph on Autopilot, the vehicle’s semi-automated driving system.

Much of the media coverage about the incident revolved around the CHP maneuver to safely bring the vehicle to a stop. The officers were able to manipulate Tesla Autopilot to slow down and ultimately stop mid-highway using two patrol vehicles, one in front and one behind the ‘driverless’ car.

But USC Law Professor Orin Kerr mused online about a constitutional quandary relating to the stop, asking, “At what point is a driver asleep in an electric car that is on autopilot “seized” by the police slowing down and stopping the car by getting in front of it?” This question centered around when a person asleep was seized,a reasonable 4th Amendment inquiry given the U.S. Supreme Court standard that a seizure occurs when a reasonable person would not have felt ‘free to leave’ or otherwise terminate the encounter with law enforcement.[1] 

Kerr’s issue was largely hypothetical given that the police in this situation unquestionably had the legal right to stop the vehicle (and thereby seize the driver) based on public safety concerns alone.

However, a larger 4th Amendment question regarding semi-automated vehicles looms. Namely, what constitutes’reasonable suspicion’ to stop the driver of a vehicle on Autopilot for a traditional traffic violation like ‘reckless driving’ or ‘careless driving’?[2] Though there are no current laws that prescribe the safe operation of a semi-autonomous vehicle, many common traffic offenses are implicated by the use of automated driving features.

Some ‘automated’ traffic violations will be unchanged from the perspective of law enforcement. For example, if a vehicle on Autopilot[3] fails to properly stay inits lane, the officer can assess the vehicle’s behavior objectively and ticket the driver who is ultimately responsible for safe operation of the automobile.Other specific traffic violations will also be clear-cut. New York, for example still requires by statute that a driver keep at least one hand on the wheel.[4] Many states ban texting while driving, which though often ambiguous, allows for more obvious visual cues for an officer to assess.

However, other traffic violations like reckless driving[5] will be more difficult to assess in the context of semi-automated driving.

YouTube is filled with videos of people playing cards, dancing, and doing various other non-driving activities in their Teslas while Autopilot is activated. While most of these videos are performative, real-world scenarios are commonplace. Indeed, for many consumers, the entire point of having a semi-autonomous driving system is to enable safe multi-tasking while behind the wheel.

Take for example, the Tesla driver who is seen biting into a cheeseburger with both hands on the sandwich (and no hands on the wheel). Is this sufficient for an officer to stop a driver for careless driving?Or what about a driver writing a note on a piece of paper in the center console while talking on the phone. If during this activity, the driver’s eyes are off the road for 3-4 seconds, is there reasonable suspicion of ‘reckless driving’that would justify a stop? 5-6 seconds? 10? 20?

In these types of cases, the driver may argue that they were safely monitoring their semi-automated vehicle within the appropriate technological parameters. If a vehicle is maintaining a safe speed and lane-keeping on a low traffic highway, drivers will protest – how can they be judged as ‘careless’ or ‘reckless’ for light multi-tasking or brief recreation while the car drives itself?

The 4th Amendment calculus will be especially complicated for officers given that they will be unable to determine from their vantage point whether a semi-autonomous system is even activated. Autopilot is an optional upgrade for Tesla vehicles and vehicles that are equipped with L2/L3 systems will often be driven inattentively without the ‘driverless’ feature enabled. Moreover, most vehicles driven today don’t even have advanced automated driving features.

A Tesla driver whose hands are off the steering wheel could be safely multi-tasking using Autopilot. But they could also be steering with their legs or not at all. This leaves the officer, tasked with monitoring safe driving for public protection, in a difficult situation. It also leaves drivers, who take advantage of semi-automated systems, vulnerable to traffic stops that are arguably unnecessary and burdensome.

Of course, a driver may succeed in convincing a patrol office not to issue a ticket by explaining their carefully considered use of the semi-automated vehicle. Or the driver could have a ‘careless driving’ ticket dismissed in court using the rational of safely using the technology. But once a police-citizen interaction is initiated, the stakes are high.

Designing a semi-automated vehicle that defines the parameters of safe driving is complex. Crafting constitutional jurisprudence that defines the parameters police behavior may be even more complex. Hopefully the Courts are up to the task of navigating this challenging legal terrain.

Jesse Halfon is an attorney in Dykema’s Automotive and Products Liability practice group and a member of its Mobility and Advanced Transportation Team.


[1] United States v. Mendenhall, 446 U.S. 544, 554 (1980); United States v. Drayton, 536 U.S. 194, 202 (2002);Florida v. Bostick, 501 U.S. 429, 435-36 (1991).

[2] Some traffic violations are misdemeanors or felonies. To make an arrest in public for a misdemeanor, an officer needs probable cause and the crime must have occurred in the officer’s presence.  For a Terry stop involving a traffic misdemeanor, only reasonable suspicion is required.

[3] Tesla Autopilot is one of several semi-automated systems currently on the market. Others,including Cadillac Super Cruise Mercedes-Benz Drive Pilot and Volvo’s Pilot Assist offer comparable capabilities.

[4] New York Vehicle and Traffic Law § 1226.

[5] Most states have a criminal offense for reckless driving. Michigan’s statute is representative and defines reckless driving as the operation of a vehicle “in willful or wanton disregard for the safety of persons or property”.  See Michigan Motor Vehicle Code § 257.626. Michigan also has a civil infraction for careless driving that is violated when a vehicle is operated in a ‘careless or negligent manner’. See Michigan Motor Vehicle Code § 257.626b

The Battle For the Curb

Recently, Kevin wrote about how CAVs could alter the shape of cities. While CAV deployment is still in its infancy, the boom in ride sharing is already changing the design of cities. In Washington, D.C. the city government has announced the creation of five pickup and drop off zones that are reserved for ride shares 24 hours a day. The zones are also used for commercial loading and unloading, and are located near highly trafficked areas.

The creation of these zones in D.C. are part of a greater discussion of how cities use the curb. Right now, there is a lot of competition for the curb, from parking meters, to bike lanes, to drop off zones like the ones in D.C. And companies like Coord have started to keep track of everything that is going on near the curb, with an intent to build out a database that can be used by city planners and anyone else interested in what’s happening at street level. Any changes that CAVs make to cities will no doubt start at the curb – which means city governments need to figure out just what’s going to happen on the curb. Will cities be willing to give up their venue from street parking? Or will a boom in AVs cause that revenue to disappear on its own?

I’ve written in recent weeks about the impact of autonomous vehicles on city design. Choices made by both city planners and CAV operators in the coming decades will play key roles in determining whether our new transportation paradigm is one of compact, walkable cityscapes that accommodate traffic of all sorts, or one that spurs increased suburban and exurban sprawl and is truly designed only with car transport in mind. One particularly important aspect of this question is to what extent CAVs will integrate with current mass transit rather than attempt to replace it.

Some companies, such as Ann Arbor based May Mobility,are purposely seeking out opportunities to integrate with local transit. The company recently contracted with Columbus, OH to begin operating their CAVs on a short loop through downtown along the Scioto River. A small-scale project like this has the potential to improve traffic flow in the central city without incentivizing people to move ever farther away from the urban core. A deal was also announced between May and the state of Rhode Island to run autonomous shuttles that will connect public transit lines in the nearby cities of Providence and Olneyville.

Its certainly possible that May’s long-term ambitions are bigger. They may hope to use their autonomous technology to compete with companies like Ford, Waymo and Uber to provide people with a primary mode of transportation. For now though, services like this should be viewed as a model for cities seeking to promote vibrant urban centers.

Many cities across the country, even those without longstanding strength in public transit, have already committed serious resources to revitalizing and maintaining their urban cores. Kansas City is planning a roughly $300 million extension of an existing light rail line. Phoenix, which first opened its light rail line ten years ago, passed a ballot initiative in 2015 to raise new funds fora 66-mile expansion of the system. And public transit is not supported only by public money. In South Florida, a privately owned, high speed commuter train recently opened to carry passengers between Miami, Fort Lauderdale, and West Palm Beach.

Cities investing so heavily in large-scale public transit certainly have a demonstrated interest in the economic development that comes with urban revitalization. Furthermore, they see transportation as a key factor in spurring renewed growth. If these localities are not careful though, they may see their careful plans laid to waste by the onset of CAVs. In the post-WWII era, the dominance of the automobile contributed to the emptying out of city centers and the paving over of vast swaths of land. Looking ahead, it’snot hard to see the rise of this new technology thwarting the plans of the most well-intentioned cities.

Those that hope to back up their commitment to public transit and sustainable living will need to think carefully about how transportation technologies should be accommodated. For now, the May Mobility model may be attractive for its intentional compatibility with other forms of transit. Looking ahead, as CAVs become more advanced, such companies will likely move to take over more of the transportation market. Cities need to be aware of that possibility and consider how to design their infrastructure and transportation policies to integrate CAVs into existing plans, lest they betaken over by them.

California has become the second state in the nation to permit connected and automated vehicles (CAVs) to operate on public roads without a safety driver. With the recent announcement that Waymo has obtained approval to test driverless CAVs in a handful of Northern California communities, the state joins Arizona on the leading edge of the driverless vehicle revolution. Similar to the Arizona experiment, which I wrote about recently, California has positioned itself to play a key role in shaping the speed and direction of growth in the CAV industry.

California’s regulatory apparatus, while not without its critics, will provide an interesting contrast to the relatively lax system enacted by Arizona. The remainder of this post will explore the key differences between these two approaches to governance of CAVs.

Arizona requires merely that CAV operators submit written confirmation to the State that each vehicle complies with all relevant federal law, that it is capable of reaching a “minimal risk condition” when necessary, and that it be capable of complying with traffic and safety laws. California, by contrast, has a handful of more specific requirements. In addition to the need to comply with federal law, California requires that driverless CAV operators:

  • Notify local authorities in communities where testing will take place
  • Submit a law enforcement interaction plan
  • Certify that the vehicles meet the autonomous vehicle Level 4 or 5 definition of the Society for Automotive Engineers
  • Maintain a communication link between the vehicle and a remote operator
  • Inform the DMV of the intended operational design domain
  • Submit an annual disengagement report to the DMV
  • Submit collision reports to the DMV within 10 days of a crash

In addition to these requirements for driverless vehicle testing, California has a further set of requirements before driverless CAVs can be certified for public use. This supplemental set of requirements generally revolves around data recording and security against cyber-attacks.

Critics have argued that even California’s approach to CAVs is not safety conscious enough. Consumer Watchdog, a California public interest group, has raised an alarm that the state is merely taking Waymo’s word that it has met requirements “without any real verification.” The organization has also suggested that California’s regulations are not substantively strict enough, arguing that they are turning people “into human guinea pigs for testing [Waymo’s] robot cars.” Proponents though, argue that the safety concerns are overblown in light of the potential for vast improvements relative to error-prone human drivers.

While the debate over how much regulation is proper persists, it is notable how quickly California seems to be following in the footsteps of Arizona in the rollout of CAVs. One key argument in favor of Arizona’s light touch regulation is that it has positioned the state to take the lead in development of this new technology. California’s oversight, while not enough for some, is undoubtedly more rigorous than that of its neighbor to the east. Arizona’s approach does appear to have given the state a short head start in the CAV race. California’s progress, though, indicates that modest increases in state oversight may not present a substantial barrier to the adoption of this new technology.