Next Generation Transport: Self-Driving Vehicles

Cars are more than just an item of luxury, their primary use is to transport persons and goods from one point to another. With good roads and affordable fuels, vehicles connect cities and further the expansion of communities. These in turn improve the economy and provide jobs for citizens. Yet, there is one job that self-driving cars might be taking away: drivers.

From America to Asia, companies are testing vehicles that require little to no driver interaction. Self-driving cars, driverless cars, or highly automated vehicles (HAVs), as they are called, would have a tremendous impact on the world’s economy.

What is a Self-Driving Vehicle

Highly Automated Vehicles, HAVs, as they are known, are vehicles that drive themselves some or all the time. It means even with the driver seated, the vehicle can operate on its own with or without much human input. HAV, therefore, represents a spectrum of vehicles, leading to the formulation of an international standard known as the J3016.

Created by the Society of Automotive Engineers, the J3016 international standard identifies 6 levels of HAVs. These levels depend on the level of automation in the vehicle. The J3016 standards state that you are driving the vehicle if you have to accelerate, brake, and steer a motor vehicle.

The first level is 0 (zero), which represents no automation at all. These are the old model vehicles which require constant driver action for the vehicle to work. Levels 1 and 2 make way for some form of automation. For example, you could leave your vehicle in cruise control and the vehicle would keep moving without much work on your part. Even then, you would need to steer, brake, or accelerate to keep the vehicle moving the way you want it to. In all, levels 0 through 3 have major driver support features like the ones mentioned above.

Levels 3 through 5 allow for more automation. 3 and four are lower levels, though, and vehicles with such specifications can drive on their own only under specific conditions. SAE level 3 vehicles, for example, can self-drive in a traffic jam. The sensors allow the vehicle to maintain a regular distance from other vehicles in the jam and keep a steady pace as and when needed. SAE 4 cars might not even have steering wheels, meaning there is little need for a driver in the driver’s seat.

SAE 5 vehicles are the true driverless vehicles. They can drive under any condition and on any kind of road around the world without much trouble.

Who Has a Self-Driving Vehicle?

Waymo, a subsidiary of Google’s parent company, Alphabet, runs a commercial autonomous vehicle service in Phoenix since December 2018. Waymo offers a ride-hailing service to a select group of riders, the service operates on roads within a 100 square mile, Wired reported. While these are robo-taxis and autonomous, a human safety operator sits behind the wheels during the journey.

Lyft and Uber are both working to put out driverless vehicles for commercial use. Uber, for example, spent more than $450 million in R&D in autonomous vehicles (and other related vehicles) in 2018 alone, TechCrunch reported. Lyft is also offering self-driving services to a limited group of riders as part of its continuous research into the field. The ride-sharing company has a fleet of 30 cars and has been in operation since May 2018.

Tesla is looking to deploy its own autonomous vehicles in 2020. In the meantime, however, Elon Musk’s company is promising a $5,000 feature on its Model 3 vehicles that would offer a ‘full self-driving’ service, amidst scorn by industry experts.

Many companies are also developing self-driving trucks for the logistics industries around the world. Amazon is relying on one such built by Embark to distribute cargo, according to reports. These vehicles have drivers behind the wheel, however, in case a human instinct is needed. In February, 2019, Embark tested a driverless truck journey from California to Florida with minimal human intervention. Also, both Mercedes and BMW are investing in autonomous vehicles.

All the Doubts

Majority of car accidents are a direct result of human error. From drunk driving to overspeeding, the human factor has proven to be prone to more errors than any other auto malfunction. Absence of up-to-date reporting makes it difficult to ascertain the exact number of accident victims in some African countries. Conservative reports, however, indicate that Africa records an accident death rate of 24.1 per 100,000 people. In Ghana alone, fatalities increased from 1,802 in 2015 to 2,084 in 2016, according to data from the National Road Safety Commission.

94% of road accidents in the USA are the result of human errors, while the figure is 75% in the UK, a risk analysis report indicated. By removing the human driver from the vehicle, it is hoped that this would cut car crashes down significantly.

Having said that, artificial intelligence-powered vehicles are so far not so safe either. In May 2018, a self-driving Uber vehicle crashed into a pedestrian during a test drive in Arizona. A few months before that, a semi-autonomous Tesla Model X rammed into a concrete barrier and killed a software engineer.

Testing AI

AI-powered vehicles still need to be taught what is an obstacle and what is a minor patch of raised earth. To do that, manufacturers of self-driving vehicles involve the services of test drivers to monitor and make notes of all the unusual patterns that their vehicles follow. But that could lead to a myriad of different results, as different drivers have unique driving patterns. That is where another form of AI is teaching the self-driving cars to drive themselves.

Waymo, arguably the leader in self-driving vehicle technology, uses a computerized version of American streets to test their vehicles. The vehicles then drive through these digitalized streets repeatedly to make them better and smarter. These vehicles are clocking thousands of miles in order to provide more accurate data about their driving ‘behaviours’ for research purposes.

According to the RAND Corporation, such vehicles need to have up to 11 billion miles of travel time to have enough data in order to create any reliable statistics.

But waiting to have a near-perfect HAV would lead to more casualties, another RAND report indicated. How safe should HAV be before rollout? Should they be 75% safer or is 10% safer just enough? From now till when we deem driverless vehicles slightly safer, many people could die, making the pursuit of perfection a near-enemy of what is already available.

According to the Society of Risk Analysis, however, people are willing to ride in driverless cars if it would reduce crashes by at least 75%. While HAVs could bring down the number of car crashes with time, other risk factors arise as well. For instance, cybercriminals could cause fatal crashes if they are able to hack into self-driving vehicles.

Costing AI

Automated vehicles are more costly to produce than traditional vehicle models. The sensors, radars, and other integrated communication devices add as many dollars to create as they do functionality.

As an example, take Lidar, a detection system that measures distance by shining laser lights on surrounding objects and calculating the reflected pulses on these objects. This sensor is critical for the functionality of any highly automated vehicle that will not rely on human driving. One Lidar sensor cost about $75,000 when it was first made by Alphabet’s Waymo. While that has fallen significantly to just about $5,000 per unit, the effect on the overall price of a self-driving car is enormous. The effect of this on adoption rate could be substantially negative.

But this could be a small figure compared to the cost of accidents. In Ghana, road accidents cost the state at least $230 million a year; over 2,300 lives were lost in 2018 alone. This does not include the effect on livelihoods when a bread-winner passes, which could take the cost closer to the billion figure each year. In America, traffic accidents cost the nation $871 billion a year in direct and indirect effects.

If highly automated vehicles (HAVs) will ameliorate the spate of road accidents, though, the poor will have to be able to afford these vehicles as much as the rich. One way to work around the affordability hurdle is to expand mobility-as-a-service platforms. Through ride-hailing companies like Uber, low income people could still access safer vehicles via ride-sharing apps. Large corporations operating robotaxis would offer a safer, more accessible option to the SAE 3 and 4 vehicles as well.

For these ride-hailing companies, eliminating drivers could reduce their costs greatly. An example of how much drivers add up to the cost of ride-sharing can be seen in Uber’s 2017 figures. Uber made an estimated $37 billion in American fares in 2017. Out of that, $30 billion went to drivers alone. It therefore stands to reason that there’s a potential for driverless cars to reduce the cost of access to poor income households in general.

Legislation

In order for automated vehicles to become commercial, legislation needs to be strategically formulated. These regulations would have to cater to both the safety aspect as well as cost of operation. Questions about ownership and government subsidies for the development and adoption of driverless vehicles need to be addressed in many countries before self-driving goes mainstream.

Nestled within the safety concern are the implications of cybercrimes and terrorism within the larger conversation of national security. With the vulnerability of internet software and hardware, hacking into autonomous vehicles on a large scale could prove catastrophic, the fear of which would affect and limit the roll-out of these vehicles.

Data privacy laws would need to be beefed up to delineate what companies can collect and how they can use it as well. Also, vehicle accessibility across countries and states would need a streamlining of regulations across these nations. Regarding these, legislation would be needed to cover network services as well as the standardization of automated vehicles. For countries like America, this could lead to discrepancies within state and federal governments. The National Conference of State Legislatures currently holds a database of all the legislation currently written concerning self-driving vehicles.

Legislation, no doubt, hopes to manage risk for the individual, the country, and the shareholders. As a result, the phasing in of automated vehicles would require a special kind of laws. Which roads, for example, are going to use automated-vehicles only and which ones would have mixed traffic? And, as human-driven vehicles are phased out, will there be a higher user fee in favour of self-driving cars?

All of these are possible pain points that have to be addressed comprehensively before the first commercial self-driving vehicles will hit the streets.

Conclusions

The driverless vehicle economy could be worth $7 trillion by 2050, a 2017 Intel report indicated. For starters, driverless cars could, with time, become cheaper to maintain than human-driven cars. With AI comes cleaner fuels too, meaning the environment could become better, all things remaining equal.

By encouraging ride-sharing to reduce costs, traffic congestion could fall significantly. The absence of drivers could also cut down road crashes caused by human error, making our roads cleaner, safer, and more efficient. At the same time, smart vehicles would contribute to smarter cities. Efficiencies would cut across all aspects of the economy, from easier navigation systems, shorter rides, and better waste management systems. Communication systems would also be better in a world where majority of the cars don’t need drivers.

However, HAVs need to be safe, soon.