Individualist Auto Culture: Gaps at the Heart of Autonomous Vehicle Safety
The technological and regulatory basis for autonomous vehicle (AV) development will cement modern-day automobile individualism into our AV future.
Modern AVs are complex and unique machines, but the functions they need to perform are very familiar to the human driver - navigate roadways from point to point, avoiding obstacles and other vehicles while complying with laws governing the movement of traffic. To do this, the overwhelming majority of AVs rely on input from a series of sensors including cameras, LiDAR, and radar, integrated by a computer to create a view of the surrounding environment. While they are sometimes capable of communication from vehicle to vehicle across fleets from a similar manufacturer, most AVs do not incorporate secondary mechanisms to detect the positioning of neighboring vehicles. And while early attempts at AV development often placed an emphasis on the role of surrounding infrastructure in interacting with the vehicle, more-recent developments have largely left this concept behind, leaving gaps in the deployment of AV-oriented smart infrastructure. By and large, each AV operates as a singular unit, making its way through an unpredictable environment using the sensors and computing power immediately at hand.
AV proponents describe a wide range of supposed benefits to AV deployment, most notably their improved safety over conventional autos. Now, the central claim about safety by proponents of AVs - that they are substantially safer than ordinary cars driven by fallible humans - is demonstrably true. In this way, the push for widespread adoption of AV technology marks a vast improvement over advocacy for technologies like the “flying car”, with safety implications so glaringly out of control that not even the slickest-looking vehicles pass a basic smell test. Still, the persistence of ideas like the flying car in film, the popular imagination, and at the radical fringe of transportation technology development speaks to our overwhelming urge to assert the individual over the societal in transportation, even at the expense of safety. Discouragingly, recent high-profile collisions involving AVs illustrate how their technological development has fallen prey to these same impulses.
The safety concerns AVs are intended to address are wide-ranging, but one thing they hold in common is that they are underpinned by our free-form, individualist car culture. Consider these all-too-familiar scenarios: A poorly-trained teenager is too inexperienced to negotiate a tricky turn. A neglectful owner fails to change a bald tire and loses control after it violently goes flat. An elderly man reluctant to give up his keys cannot react in time to stop for a pedestrian in a crosswalk. We attribute each of these failures - in training, maintenance, and operating fitness - to individuals. The systemic, widespread nature of similar failures around the world, year after year, goes under-acknowledged or met with a helpless shrug by leaders and policymakers.
Fortunately, a well-designed AV might be able to mitigate or eliminate the risks posed by these scenarios. Still, while they are safe and reliable under many circumstances, it is clear that when AV failures happen, it is often because their technology replicates the same issues that beset human drivers - albeit on a lesser scale. A pedestrian making a mid-block crossing at night is struck and killed when the vehicle’s sensors cannot identify her. A man dies after his Tesla cannot make out a tractor-trailer against a bright-lit sky. When we talk about fatal collisions like these as a result of an AV’s failure to “see” in the dark or in the sun, as news articles often do, a very revealing act of anthropomorphization takes place. Suddenly, in this framing of vehicle-as-individual, it becomes very challenging to view these events as the cumulative result of many deliberate cultural, political, and technological choices. If the best an AV can do is carefully mimic the performance of a human driver - even an excellent one - it will be a long time before this new technology provides the miraculous level of safety performance that advocates have promised.
A central irony in the discussion around AV safety - and the feasibility of widespread AV adoption as a whole - is that many of these concerns have already been effectively addressed by other transportation modes like passenger rail, freight rail, and aviation. In fact, the basis for safe movement via these “old reliable” modes was frequently established decades ago, and has been hiding in plain sight as a source of inspiration for AV development; both rail and aviation have already achieved advanced forms of driverless operation through technologies like the automated people mover and aircraft autopilot. The success of these modes has depended on a radically different way of thinking about movement from the individualist world of autos.
Take a freight train traveling across the west from Chicago to Los Angeles: It faces many of the same challenges that an AV might face while making the trip - weather variations, competing traffic, difficult terrain - yet it navigates these challenges in a dramatically different manner. Though both the tracks and other trains are privately built, operated, and maintained, for example, Federal Railroad Administration regulations closely govern the design of the rails, inspection frequencies for key equipment, training qualifications and service rules for operators, and the functionality of signal systems and warning devices. Even in an entirely private transportation environment, a program of standard technologies, shared and “intelligent” infrastructure, and interoperable communications facilitated by a government regulator allows a train to move safely and confidently from point to point. The AV, meanwhile, can rely on no such comprehensive safety framework to shape its environment and operational performance, depending only on the inputs from its sensors and whatever programming it has received to navigate potholes, snow squalls, congestion from other cars, and a varied landscape of signage and road markings.
Modern aviation, long considered the gold standard in safety performance across commonly-traveled modes, takes many of these cultural safety imperatives even further. The Federal Aviation Administration, for example, not only monitors the development of new aircraft, but directly certifies new aircraft designs for airworthiness. Carriers are required to systematically identify, document, track, and mitigate safety hazards. All accidents are investigated by a national-level body of transportation experts, with results and findings reported to the industry so safety performance can be continuously improved. Certainly no such measures are in place for the average beater in the average driver’s garage - nor are they fully represented in the great morass of state and federal rules and guidance surrounding AVs.
The automobile landscape, upon which AV technology will be superimposed, is unlikely to be transformed by any such radical safety framework anytime soon. Some regulations and requirements exist, but they are a patchwork, nowhere near as comprehensive as those for other modes. Governments, either cowed by industry or reluctant to stifle experimentation, have failed to establish a cohesive set of regulatory requirements. AV firms themselves have little incentive to collaborate with one another in a world of proprietary technologies and what are assumed to be vast untapped profits. At the root, though, and most importantly, every stakeholder in the discussion is encumbered by a century’s-worth of inertia around our conception of roads and cars - that the street is a blank slate meant for private vehicles to complete private trips with as little interference as possible, for better or worse.
delorme conducts audits and inspections to improve safety, operations, maintenance, and management practices at a large east coast transit agency