A few weeks ago, while most of the country was entombed in a deep freeze from a freakishly cold winter, I found myself in sunny south Florida in the middle of a traffic jam. As is typical of most traffic jams, the repetitive actions of speeding up and slowing down, touching the brake and easing on the gas are all too familiar to those who have ever turned a steering wheel. However, this particular traffic experience ended up being the most unique traffic encounter I have ever had. You see, while I was driving a brand new luxury SUV that, from the outside, looked every bit like every other similarly sized SUV sausaged into the tight Florida roadways, on the inside, something remarkable was happening. That “something” was the fact that the entire time I was behind the wheel as we negotiated the traffic, I never touched the brake or the gas pedal. Instead, the vehicle was utilizing its active cruise control technology to regulate speed and other technologies to stop the vehicle, if necessary. In essence, without me having to steer, the vehicle basically drove itself.
Does this seem farfetched or like a futurespeak feature for a far off time? Think again, as the technologies I experienced are sitting as close as your nearest automotive dealership— and not from just one manufacturer. Vehicles with self-piloting capabilities are beginning to populate our roadways, so what does this mean for law enforcement and the fundamental way in which we do our jobs?
The idea of an autonomous automobile isn’t something new. In fact, the most notable project promoting the idea of a driverless car, spearheaded by Google, has been underway for almost five years, with more than 100 cars on the roadway being tested. While that car is designed to be 100% autonomous in nature, it is important to point out that production vehicles like the one I drove in Florida are partially autonomous in that, in certain circumstances, the vehicle is capable of intervening and operating on its own. Those systems can be temporarily disabled and the rest of the time, the vehicle functions like a traditional car. It’s also important to note that these systems are presented as “driver aids” and not “safety devices.” This is because the inherent issues of relying on a computer system to replace the driver in all circumstances are perilous, especially when relying on it to be 100% accurate, 100% of the time. Anything less would be unacceptable and, perhaps, contribute to creating a larger problem than the one it’s intended to solve. So, for now, autonomous intervention, even to the level that the systems will actually steer a vehicle around an object, as found in emerging technologies like Ford’s Smart Obstacle Avoidance Technology, Volvo’s Pedestrian and Cyclist Detection and Chrysler’s Forward Collision Warning will remain in the background until needed. Still, the wave of progression in this area is growing, so what are some potential areas where it could affect our daily job environment?
At the root of all traffic law is the requirements for a vehicle and an operator. When looking at Arizona Revised Statutes, ARS section 28-3151 reads: “A. Unless exempt pursuant to this chapter, a person shall not drive a motor vehicle or vehicle combination on a highway without a valid driver’s license and proper endorsement as prescribed by this chapter.” Such verbiage is fairly consistent among states, so the enforcement of such a statute, something we do every day, could become very muddled with autonomous operation. This is because, if the vehicle is operating on its own at certain times, is it really being “driven” or is the occupant in the driver’s seat just a passenger? Furthermore, if the system takes over under certain conditions, would that make the requirement for a driver’s license a partial one too?
Consider a situation where you stop a driver for a traffic infraction that was caused, not by driver input, but by system input. Who gets cited, the driver or the car? How would you know the difference? You could argue that the driver is ultimately responsible, but if the vehicle does what it wants to, when it wants to, how can a court assign the driver at fault when they have no control over the system’s decision process or behavior? Since the system, at this point, can be deactivated by the driver, and is situational, who’s to say a driver couldn’t state the system created the infraction rather than them? Does the vehicle itself become self-policing and record infractions on its own? If you remember the movie, The Fifth Element with Bruce Willis, that is exactly what played out for drivers in that scifi world. Are we headed that way and, if so, does that reduce or eliminate the need for a traffic division in a department? So, the reality of drivers being removed from the driving process does pose some licensing and traffic enforcement quandaries.
I mentioned these systems are considered driver aids, which brings up some interesting questions regarding crash investigation and causal factors. We have already seen the effects of dynamic vehicle control systems in crash investigation as the application of stability control, traction control, hydraulic and dynamic braking assist, ABS and other systems can dramatically affect roadway markings, vehicle behaviors and causal relationships in an event. The days of measuring wheel lockup, yaw marks and co-efficient of drag alone to figure out what happened are pretty much gone, so how do we prepare for vehicles that stop and steer themselves? As an example, if an autonomous system decides to slow a vehicle abruptly and steer around an object to the left, and that action causes the vehicle to strike an approaching vehicle head-on because the system couldn’t see it, who is to blame? The system? The driver for not intervening? The manufacturer? What about situations where a driver is involved in a crash that they state could have been avoided if they had been in control? Who do you believe? Roadway evidence will still determine the path of the vehicle, AOI, etc., but where do you go from there? Do we install cameras in all cars that monitor all activity in and out of the car for post-crash analysis and assignment of responsibilities? Does that mean the auto manufacturer will be required to testify and present witness to every crash that goes to litigation? So, while the greatest percentage of crashes are low speed property crashes that are typically rear-striking where inattention to driving is the cause, there may be a place for autonomous technologies. However, the task of investigating higher level crashes could become significantly more complex.
As a former DUI enforcement officer, the idea of “physical control” was a big part of meeting the requirements of revised statute for such a charge. In fact, it was central to the case. So, with autonomous technologies, is the “driver” of a vehicle that’s operating in self-driving mode really in physical control if they have no influence on vehicle actions? Does this relegate them to being a “passenger” at that point, thereby neutralizing their culpability? While physical possession of the keys and proximity to the ability to operate a vehicle can be factors in traditional DUI enforcement, does having a key to an autonomous vehicle end up being as relevant as having a key to your house or gym locker? What about probable cause for the stop? As an example, if you observe a vehicle having trouble maintaining a consistent position in a travel lane—a classic sign—and stop that vehicle only to have the driver maintain that the vehicle was steering itself down the roadway using lane markings, independent gas and brake, etc., would the probable cause hold up in court? If we agree that the sole reason for that stop was the observed driving behavior pre-stop, if the driver ends up being under the influence, does that constitute an invalid stop because autonomous behaviors were at play? While we all want to get DUIs off the road, we also want to make sure they “stick” in court, so the integration of autonomous systems poses a future challenge in this area.
Since a good deal of law enforcement revolves around traffic enforcement, and governmental processes revolve around vehicle statutes, a large reduction in driver culpability and requirements could mean a large reduction in officers and governmental services like the DMV. Less crashes, DUIs and traffic infractions could mean a significant change to the landscape of the typical law enforcement agency. It could also mean the development of forensic analysts that specialize in autonomous vehicle behaviors. In other words, the jury is still out on a lot of this, but it definitely is coming, so the law enforcement world will have to evolve along with the technology to meet these new investigative requirements.
If you recall at the beginning of this article, I didn’t mention the manufacturer of the vehicle I was driving. I didn’t omit it because the technology didn’t work, because it did—very effectively. I omitted it due to my own mistrust of handing over my livelihood to a computer as we hurtled toward stopped traffic at 45 mph with the impending requirement to stop. This thought was bolstered by insufficient time with the system to completely trust it, and that goes for any system by any manufacturer. It was also bolstered by knowing some of the limitations of these systems. For example, the Google cars, much like all active lane guidance systems, don’t work in snow or areas where there are no definitive lane markings. The same applies for active cruise control systems in that, in order to work properly, there must be another moving “object” in front of them for the forward sensing radar to maintain a proper following distance when necessary. If that object disappears, then the system may interpret open road ahead and that can cause it to accelerate to reach the maximum speed the driver has programmed, which may or may not be appropriate.
Another acknowledged limitation of autonomous systems is that they can’t recognize human behaviors like hand gestures, so this can pose a problem in traffic control situations. Since drivers are required to obey the direction of an officer, it could become difficult to decipher if a driver failing to obey your command is attempting to, but the vehicle won’t let them, or if they are just being inattentive or clueless. Another issue is that a fully autonomous system is designed around roadway mapping, so it’s only as good as the maps it knows. So, if there is an error in the maps, or a new roadway is built that isn’t on the computer’s map, this could lead to some bizarre driving behaviors as vehicles either turn where there is no road, or don’t turn when there is. This scenario is a bit further out, but something that needs to be solved.
As an EVOC instructor and racer for more than 20 years, the idea of a selfdriving vehicle goes against pretty much everything my career stands for. But as with all environments, we must learn to adapt or perish. The automotive landscape has changed profoundly in the last 10 years. We have electric vehicles as mainstream buying choices, and even as law enforcement vehicles. Our patrol vehicles have sophisticated dynamic chassis and traction control systems. There are data capturing devices integrated into every aspect of vehicle operation. It was only a matter of time before computer assisted dynamic vehicle control would begin to transition from a passive to an active role.
For now, the technology remains as a driver’s aid, and I have to admit that it was pretty intriguing to drive through traffic successfully without touching either the brake or gas pedal. I will also admit that there are places where these systems can help address that inattention to driving factor that plagues so many causal relationships in crashes. So, in the context of today’s autonomous intervention, the systems don’t replace the driver as the primary input device, but the day may be on the horizon where steering wheels will disappear from vehicles, and our patrol cars will become officer-moving devices rather than officer-driven ones. The challenges mentioned in this article may not materialize tomorrow, next week or next year, but they are coming and we need to be ready so we don’t end up getting taken for a ride in the most undesirable way.
JP Molnar is a former state trooper and has taught EVOC since 1991 for various agencies. He has raced for for almost three decades and has taught at numerous high performance racing schools.