Self-driving vehicles are fueling significant and exciting technological development.
Self-driving vehicles are fueling significant and exciting technological development: Advanced sensors and neural networks are two examples. But the technology needed to enable autonomous driving goes beyond just the vehicles: The roads they operate on can also benefit from technological advancements to become safer and easier to use for such vehicles. This is an area that’s gaining increased prominence, though not from all stakeholders.
In the US, for example, the Department of Transportation recently published policy guidance for self-driving cars. And while it tells carmakers what to do to ensure the safety of their vehicles, it doesn’t look at what road-builders or drivers need to do. This narrow focus is understandable: Semi-autonomous cars still represent a tiny fraction of all vehicles, and even the boldest forecasts don’t envisage this balance shifting significantly for some time.
This leads us to an important question: Is it right to spend money on road changes that benefit a minority of users? And could such alterations actually make the roads less safe for human road users?
While this debate goes on, those designing and building autonomous vehicles are actively pushing for enhancements to the road networks. And certain governments are responding.
Lane Markings Need to Improve
Vehicle manufacturers suggest that a straightforward improvement would be to mark roads correctly, in accordance with governments’ own standards.
In 2016, the mayor of Los Angeles, Eric Garcetti, boldly stated that the city would be gearing up its roads for the forthcoming growth in autonomous vehicle numbers. He said: “[We] will ensure that the demands and challenges of this new technology are fully considered as we invest in our infrastructure and plan for Los Angeles’ transportation future.”
Later, as part of a media showcase, Garcetti got into a self-driving Volvo. The car had difficulty driving itself. According to Volvo North America President, Lex Kerssemakers, sitting alongside the mayor, this was because the car’s cameras couldn’t pick out the faded road markings. Kerssemakers exclaimed: “It can’t find the lane markings. You need to paint the bloody roads here.”
In this month’s In Focus, we look at the developments in the automotive electronics sector, supply/demand scenario, and how suppliers are keeping up with rising demand.
Kerssemakers is not the first autonomous vehicle maker to bemoan the condition of the roads. Tesla’s CEO Elon Musk showed reporters two sets of overlapping lane markings on a freeway near Los Angeles. The duplicate lines could cause Tesla’s autonomous cars to track the wrong markings and consequently move out of their lanes. The only way Tesla could make the road safe for its cars was to map every lane in advance, meaning the cars weren’t exclusively reliant on their sensors. “We really need better lane markings. This is crazy,” Musk said. He has given funding to a group lobbying for enhancements.
Consistency of Road Signs and Signals
An associated headache for self-driving vehicle makers is the lack of consistency when it comes to signs, signals, and markings: Cross a state or national boundary and the design and format of these key elements can change.
Beyond simply improving conventional road elements, we’re also seeing a shift towards smart solutions. A team of government researchers in Singapore recently observed: “For autonomous vehicles to work well, roads, road signs, and signals may need to be mapped or made intelligent.”
Vehicle-to-Vehicle and Vehicle-to-Infrastructure Communication
The push to create smarter roadways dovetails neatly with proposed communication standards, such as automotive LTE and Dedicated Short-Range Communications (DSRC). These technologies enable Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communication, both part of the broader vehicle-to-any-asset (V2X) umbrella.
V2V systems are similar to Traffic Alert and Collision Avoidance Systems (TCAS) used in aviation to alert car drivers to potential collisions. Following a campaign lasting over 10 years, the US National Highway Traffic Safety Administration seems close to gaining political approval for mandatory transponder-based V2V systems.
V2V and V2I technology were originally envisaged as a helper for human drivers. But both can also be extremely valuable for self-driving vehicles, complementing sensor data and sharing this with other road users to build up a better understanding of conditions. At Baidu in China, researchers have suggested further benefits. For example, the hand signals made by a police officer to a stream of oncoming traffic can’t be interpreted by a self-driving car; the human driver would need to take control to follow the instructions. Baidu has outlined how V2X technology could help overcome this issue, including by providing traffic control officers with beacons that send instructions to the traffic via V2I networks.
What Does This Mean for Human Drivers?
The push to alter roads for the benefit of autonomous vehicles could result in controversial results: We could end up with roads optimized for self-driving vehicles but that are less safe for human drivers. Take this a step further, and you could have a situation where human drivers are forbidden from driving on certain roads.
Here’s why: Autonomous vehicles have greater awareness of their surroundings than human drivers and can react faster. Consequently, roads for such vehicles could be narrower, thereby being less expensive to build and releasing valuable real estate. Cars could also travel faster and with less space between them, while V2I technology could eradicate the need for road signage and traffic lights. But all of this would rely on there being no human drivers, pedestrians, or other road users in the mix.
Let’s come back to our question of whether changing our roads for the benefit of self-driving vehicles is the right thing to do. This is likely to remain a contentious issue, to which we may never get a definitive answer. More likely is that smaller-scale alterations will happen bit by bit, without specific intervention from governments.
Imagine for a moment the year 2027. One in ten vehicles now has sophisticated self-driving features. When reviewing collision data, urban transport planners realize that a large proportion of accidents at a busy junction involve autonomous vehicles. On further investigation, it turns out that this is because the cars’ sensors can’t detect a traffic light when the sun shines at a certain angle. The authorities add a second set of lights (or move the first), which immediately cuts the collision rate. What they have done is adjust the roads for the benefit of autonomous vehicles.
What about all the other road users?
About the Author