Cal Hates Autonomy

Self driving cars are sold to us as one of the most difficult technological challenges of our day. on the surface, this does in fact seem like an accurate representation. Despite all of our flaws, it is hard to replicate a human behind the wheel. We are able to take in and process a lot of information as we move down the road, make decisions, and take action. We are not perfect and sometimes we make bad decisions. To solve this, tech companies have been working to take the driver, the one making these bad decisions, out of the picture.

The advantage of this? Increased safety as the car will be able to take in and process even more data than we could ever dream and make decisions with a well understood logic no matter the external circumstances. On top of this, we could take back time that was previously spent focused on the road to whatever it is we want. This is the top level. We can also hear arguments about how this can allow for less traffic, reduce emissions, or even allow your car to become an income producing asset. These claims while potentially valid, need more and I want to focus on the acceptance of this technology today.

Today, different companies are at different levels of autonomy. Some cars simply offer driving assist like dynamic cruise control or auto steer. Some can read signs and traffic signals and others can even fully autonomously drive you through the city as a driverless taxi. They all work largely the same, however. They have a slew of sensors around the vehicle that take in all the information we would normally get from our eyes and allow the vehicle to know where it is on the road, and more importantly, know where other cars or obstacles are as well. From this, it makes decisions on how fast to brake or accelerate or how to turn the car. Just the way a human driver would. If we were to continue down this path we would see that these cars would eventually be able to solve the edge cases that we all immediately jump to, and all seem to be a variation of the Trolley Problem. But if that is the end goal, what is really so groundbreaking? The time freedom gained by no longer having to pay attention to the road is not necessary with tools that allow us to work and communicate remotely, and even with 100% driver involvement we find time to do other things like talk on the phone or check our emails, even if the law says we should not. Many of us also forget about the options of public transportation when posed this question. And I would agree, self driving cars would be a big innovation, and likely make our roads safer but for this to truly change the world, these cars need to do more.

Luckily they can, because just as the most attentive driver in the world is limited to what information they can receive on the road so are today self driving cars. Their sensors are limited and while they may be able to react sooner or see better in the dark, we aren’t making the revolutionary jump that is required to bring about the mass adoption of self driving vehicles. Rather, we need to find a way to get these vehicles more information than a human driver could ever process by entirely new means.

This can be done by integrating all of the cars together into a system that can directly feed information to each vehicle. Allowing these self driving cars to not only see around corners in certain circumstances but also know exactly what every vehicle around it is about to do next. This system would allow for the so called high speed caravans to zip through cities and shorten travel times. This system would work well when all cars are able to work together to build a full picture. This isn’t perfect. For one, cars last a long time and are expensive, this would only work if all cars are autonomous, something likely to take decades before full adoption. Also, this ignores what is going on off the road. Many of those previously mentioned edge cases involve people who are notably not in cars. These pedestrians on the city streets are not going to be transmitting their location or what direction they are heading. However, their actions can have major impacts on how the vehicles on the road behave. To truly let the cars roam free on the streets and expect increases in road safety, we need to also develop a way for these cars to know what is going on off the road and beyond the range of their sensors.

To solve this I am proposing a simultaneous introduction of roadside sensors. These would be able to track their immediate surroundings and inform approaching vehicles of the information they need to know, be that a potential obstacle that may enter the road or allow the vehicle to see around a corner even when the onboard sensors are obstructed. These sensors could monitor animals on the side of the road who can range in size and could be easily missed by an onboard sensor. These would also be able to decipher what pedestrians are likely to enter the roadway from a group of dozens on a sidewalk. This sensor could pick up those who are walking towards the street even if the view from the road is obstructed by those walking along the edge. Once an object or person is identified it can feed it to all approaching cars and tell them when to slow down. That way when a car rounds a corner and sees two people in the intersection, it already knows they are there and is able to slow down in time rather than decide to hit the old lady or the little kid. 

Beyond this, these roadside sensors could act as transmitters, informing cars of changes in traffic patterns or how they should operate in bizarre situations like drive throughs or car washes. If we want to eventually build cars without steering wheels and still want to pick up food without getting out of our cars we will need something to successfully navigate a car through each unique restaurant design. These transmitters could guide cars into places their onboard sensors would tell them to avoid and allow these businesses to function as our on road habits change. This could also be used in the case of an accident. If something does happen on the road it can tell the cars what lane to merge into and even transmit the signal of an approaching emergency vehicle allowing all cars to open up a gap for them to drive through. 

All of this is possible with roadside sensors that work with the cars. Without these, I am afraid that we may soon find ourselves in the cabin of a driverless car stranded amidst the traffic after an accident, at a standstill as onboard sensors get overloaded at a drive through or still impacting animals as they cross the road. All while knowing if we were still behind the wheel we would know exactly what to do based on the unique circumstances we see. 

Implementing this vision will undoubtedly be a challenging endeavor, requiring substantial infrastructure investment and collaboration across industries. However, the benefits are worth the effort. Imagine highways with self-driving express lanes, where vehicles seamlessly communicate with each other, drastically reducing travel times and enhancing safety. Picture businesses using these sensors to optimize traffic flow in their drive-throughs and parking lots.

September 29th 2023