The hottest topic in automotive and mobility industries is a self-driving car. The autonomous vehicles are poised to change how we interact with cars, our mobility habits, our cities, urban planning, and generally our way of life. Tech companies, automakers, insurance companies and governments are all gearing up for the changes that self-driving cars promise. In this guide, we try to provide an introduction to self-driving cars, technology behind it, current market players and the impact it might have on various industries and private individuals.
What is a self-driving car?
A self-driving car is a general term describing a vehicle designed to travel to a destination without a need for a human to monitor the roadway or steer the vehicle. There are various degrees of autonomy that self-driving vehicles have, based on the technology used. A fully autonomous vehicle is able to navigate without human intervention. These vehicles use various technologies starting from sensors to perceive the environment they are in, to processing large amounts of data and making intelligent decisions based on the data.
To differentiate between various types of autonomy, US Department of Transportation has come up with a categorization system:
- Level 0 - No Automation: These are ordinary vehicles that we use today, a driver is in full control of the vehicle and manipulates it through steering, throttle, brake and motive power.
- Level 1 - Function Specific Automation: This type of automation is applied to a specific function of a vehicle, such brakes. Some vehicles are equipped with pre-charged brakes, where the vehicle automatically assists the driver with regaining the control over the vehicle or slows down vehicle faster than a human driver could do on his own.
- Level 2 - Combined Function Automation: This type of automation combines two or more functions of the vehicle by coordinating them. An example of this is cruise control with lane centering.
- Level 3 - Limited Self-driving Automation: These vehicles enable a driver to cede control of all safety-critical functions in specific environments but expect a driver to take over the control, if needed. Most of the current self-driving cars today are at this level.
- Level 4 - Full Automation: In this type of vehicle, a human driver just inputs the destination and is not expected to perform any driving during the entire trip.
How does a self-driving car work?
Self-driving cars combine numerous technologies to ensure smooth and effortless transportation. This requires coordination of many sensors that gather information about the environment, sophisticated computers and algorithms (and in some cases Artificial Intelligence) to process that information, making predictions about the environment and responding to changes in the environment. The core technologies used are radar sensors, optics, LIDAR, GPS and processors.
Radar sensor - These are quite basic sensors mounted on the front and back bumpers of the vehicle and help the system detect collisions, obstacles or detours. The sensors function in coordination with other units such as gyroscopes and wheel encoder.
Optics - These are basically high-powered cameras. Cameras help the vehicle read road markings and traffic signals. The technology substitutes human eye and provides numerous images that help the vehicle determine depth of field, dimensions of objects and peripheral movements.
LIDAR - Laser Illuminating Detection and Ranging is probably one of the most important technologies powering self-driving cars. It is usually mounted on the top of the vehicle, spins continuously, scans the surroundings and builds a 3D model for the vehicle. The sensor includes a laser emitter that sends a beam around the sensor up to 100 meters, after the beam bounces back to the sensor, it is gathered in a receiver and sent to the main computer.
GPS - This is quite basic tech that we have been using for many years now and is crucial to vehicles to navigate. Without GPS a vehicle wouldn’t be able to determine its destination and its current location in relation to the final destination. Several companies are building more advanced maps for self-driving cars, as they require more data than humans.
Processor - All data gathered from sensors is processed by a computer, usually specifically built for this purpose. Some of the decisions are hard coded in the processor, like stopping at a traffic light, while more sophisticated decisions can be learned through the use of AI.
There are currently no fully autonomous vehicles on the roads, as most manufacturers are in a testing mode that requires a human driver to be present.
Who is doing what?
One of the first “self-driving” vehicles is considered to be American Wonder built by Houdina Radio Control Corporation in 1925. The founder of the company Francis Houdina equipped 1925 Chandler Metropolitan Sedan with an antenna on the back of the car. He followed American Wonder with another vehicle and a transmitter. While technically it was a gigantic RC car, it showed the interest in the development of self-driving vehicles. However, the public was not always as excited about the self-driving cars and the fears that persist today of vehicles being out of our control have been well demonstrated in a 1911 movie called The Automatic Motorist.
Many companies have been working on the self-driving technology since 1990es, when such efforts were extremely expensive and not commercially viable. One of the first companies to commercially release “Autopilot” system was Tesla. This technology constantly improves through software updates but requires a human driver to be present at a wheel and take control if needed. Tesla was also one of the first auto manufacturers to debut the functionality.
Most companies work on just technology that they apply to non-autonomous vehicles. Google started a self-driving project in 2009. The project was led by a co-inventor of Google Street View, Sebastian Thrun. He also won $2 million prize in a prestigious DARPA Grand Challenge by building a robotic car. In 2014 Google revealed its first autonomous vehicle with no steering wheel, acceleration or brake pedal. The possibilities of the small cute car, assembled by Roush Enterprise with equipment from Bosch, LG, Continental and others, were revealed when it drove around a legally blind person in 2015. Google has used vehicles from Toyota, Audi and Lexus for its tests. The initial cost of the self-driving technologies, beyond the actual vehicle, was around $150,000, most important being Velodyne 64-beam laser equipped LIDAR system.
The project was separated from Google in 2016 into a separate company Waymo, under the parent company of Google, Alphabet. The company now works with Chrysler Pacifica vans and is running public tests in Phoenix, where members of public are invited to drive in the vehicle, as long as they provide feedback. A human driver is always present, just in case.
Apple, another tech giant, has been rumored to be working on a self-driving car for many years. Most of the expectations concentrated on building a full vehicle that would compete with Tesla. The company has been hiring numerous self-driving car specialists, however in 2017 in was revealed that the iPhone maker wasn’t working on an actual vehicle but just software. Tim Cook revealed that for Apple main focus is on AI. The company is testing its technology using Lexus vehicles, but nothing particularly interesting or special has been revealed about the company’s technology so far.
Another tech company that has become a heavy weight on the self-driving car market is Intel. This was made possible through its acquisition of Israeli company Mobileye for $15.3 billion. The startup has been developing self-driving technology since 1999. It has been powering Tesla up to July 2016.
Chinese tech giants have also been trying to advance their self-driving car ambitions. Baidu has become a main driver in this process. The Chinese search engine giant has built an open-source platform, Apollo that it hopes will be used by auto manufacturers, similar to a mobile OS Android of Google. Baidu already has more than 50 partners, including Volvo, NVIDIA and Microsoft.
Another big category of companies interested in the self-driving technology is ride-sharing companies, like Uber and Lyft. The companies that just 5 years ago disrupted the transportation industry will be under a threat of disruption if they don’t move quick enough on self-driving technology.
Uber started working on self-driving vehicles by partnering with Volvo and later acquiring Otto, a self-driving truck startup. It was clear at the time that the main rationale behind the acquisition was due to Otto’s self-driving technology. In early 2017, Waymo sued Uber for stealing of its trade secrets - former Google company alleged that its former employee, Anthony Levandowski that later founded Otto and was the head of Uber self-driving efforts stole Waymo technology. Jury is still out on those allegations, literally.
Another ride-hailing giant, Lyft chose a different path. It has partnered with Waymo and hopes to test its technology through own fleet. Lyft is also working with nuTonomy, another startup developing autonomous driving technology. Autonomous technology is essential to these companies and promises to cut significant costs in their operations.
Automakers were quick to get on board with the self-driving technology. Tesla is still the leader on this market and is the only one with a commercial service on the market. Most automakers prefer partnerships with tech companies - Chrysler has partnered with Waymo, Volvo with Nvidia and Autoliv, GM with Lyft (after acquiring Cruise), BMW is working with Mobileye and Intel.
Car rental companies have also become active and found their role in the fast moving market. They usually provide logistics, coverage, maintenance and facilities that tech companies lack. Avis has partnered with Waymo, Apple with Hertz. The logistical muscle of car rentals is a gateway for tech companies to put their technology on the public roads.
New technologies and partnerships are announced almost every day, as the incumbent giants partner with small startups.
Impact of self-driving cars
Along with tech and auto companies, many other players are actively engaged in self-driving car issues. As the technology promises to disrupt many markets and way of life, actors like governments and local cities are trying to engage as early as possible, so that they can steer the technology to improve the quality of life in cities.
All governments are currently behind on the issue of self-driving cars, as they lack the insight into the technology and are trying to be careful as not to rush it to public roads, fearing responsibility in case of accidents. In the US, Department of Transportation has created a document, Vehicle Guidance for Automated Vehicles. It serves to assess safety of autonomous vehicles. Most governments require a special license to test vehicles on public roads. Cities hope self-driving cars will help get rid of congestions, prevent accidents on the road and enable a better quality of life in the cities. However, it also means loss of car-related revenue, such as traffic fines. Several cities are already working on special zones for self-driving cars.
Public and private transportation companies, freight firms as well as insurance companies are also gearing up for the impact of self-driving cars. For insurance companies self-driving cars mean diminished revenue from accident insurance, but could create new potential opportunities for other insurance services like cybersecurity of the vehicles. For freight companies autonomous technology means more effective delivery of cargo and diminished costs - meaning a lot of truck drivers would be losing their jobs.
For private individuals self-driving cars mean less time and stress during driving, more safety, availability of driving to people who couldn’t drive due to physical limitations. It also means no more ownership of a car, that is parked 95% of the time. Self-driving cars will be shared between individuals and communities, similar to ride-sharing and public transportation.
The self-driving technology itself has some shortfallings that need improvement - for example, it was revealed recently that Volvo’s self-driving car technology cannot perceive kangaroos adequately, as animals make giant leaps and sensors cannot anticipate their location after they have made a jump; recognizing traffic lights that don’t work, driving where there are no clear lane markings, responding to spoken or hand commands from law enforcement officials as well as cybersecurity are still existing challenges.
However, main challenge still remains with cities not being ready for self-driving cars. There have been several cases when self-driving car was in an accident caused by a human driver, but still fears about autonomous vehicles persist. There are numerous discussions about ethics and morality of autonomous vehicles - for example, whether AV should hit one pedestrian to prevent hitting three others. Such ethical and philosophical questions have been raised many times about humans - however, most humans act instinctively in such cases, while AV will need to be programmed in advance, making such actions rational, conscious decisions. However, bottom line remains that machines are much better at driving than humans and could save millions of lives, lost to car accidents.
While Silicon Valley remains in the leading position on the self-driving technology, Asia will probably be the first one to put the actual vehicles on the road. Singapore and Japan are already working to put the vehicles on public roads. Japan’s Robot Taxi in partnership with DeNA hopes to have driverless cars by 2020 Olympic Games in Tokyo. Singapore’s government is working with nuTonomy, a driverless taxi startup and is expected to have public on-demand busses within 2 years. China is expected to be the largest market for self-driving cars, according to Boston Consulting Group. The Chinese public is also more receptive to autonomous vehicles, with 75% saying they would ride in one. Chinese companies continue to push for more tech partners from Silicon Valley who are also desperate for testing grounds.
While the US is ahead in technology, it is falling behind on regulations. Only 23 states have currently legislation on self-driving cars (and almost all of them are quite basic). US-based companies like Google, Lyft, Uber or Apple have a hard time scaling their operations on the national level.
Such limitations require more coordination between companies across borders. While partnerships are forming every day, most parties involved look at each other as frenemies - partners that are required to implement the vision of the company but will be abandoned at the first possibility.
The self-driving technology will have numerous positive implications on quality of life and safety, it will increase our transportation efficiency and lower cost for many logistics companies. Tech companies, as well as most automakers seem to be ready to ship the technology on public roads and fully implement it within next 5-10 years. Improving the technology can only be done through testing and acquisition of data.
But this also creates headaches for governments, similar to any disruptive technology that threatens working class employees. Self-driving technology already creates thousands of jobs, but most of them are for high-skilled personnel, while people like truck drivers might be losing their jobs within next 5 years. The idea of universal income, as well taxes on AI, has been floated by Elon Musk and Bill Gates, but those ideas don’t seem to be very scalable in most countries.
The self-driving cars are no longer a thing of a distant future but a fast approaching inevitability that requires coordination of many parties, but most importantly more engagement from governments and cities to be adequately prepared for the disruption.