How Autonomous Vehicles are going to change the world
For a summarized version, click here.
Introduction
Autonomous vehicles are a very exciting and promising topic, because they will have a huge impact on society as a whole unlike any other technology.
It will fundamentally restructure the way we think about transportation and make life easier for billions of people!
The global autonomous vehicle market size is projected to grow tenfold between 2019 and 2026; from $54.23 billion to $556.67 billion in just 7 years!
The topic is not without complexity and some challenges that must be overcome, but it is being worked on heavily and we can already see a lot of progress.
Today we will discuss the history of AVs, its technology, the market, its benefits and challenges, its applications, and what a day in the life of a utopia would look like.
Let’s dive into it.
History
Although developments in the space of autonomous vehicles started in 70’s/80’s already, the breakthroughs are quite recent.
- 1977: Japan’s Trsukuba Mechanical Engineering Laboratory develops the first semi-automated car with cameras, analog computer, marked streets and a rail
- 1980: Military organizations like DARPA followed and included obstacle avoidance
- 1995: DARPA and the Carnegie Mellon University NavLab create an autonomous coastal drive on West Coast (4500 km)
- 1995+: Automated vehicle research has been primarily funded by DARPA, the US Army and the US Navy with incremental advances
- 2012: The University of Texas CS designed smart intersections for autonomous cars
- 2014+: Tesla build cars with autopilot which the driver can use but is still liable
- 2015: Delphie breaks the record set in 1995
- 2015: 6 US states including CA allow for testing of self-driving cars on public roads
- 2015/16: First trials in the UK, France and Switzerland (PostAuto in Sion, Swisscom in Zurich)
- 2016+: EU provides funding for automated driving
- 2017/18: Arizona State University develops intersection and intersection management technique (network)
- 2017: Wayme joins the race
- 2018: MIT develops car that can navigate unmapped roads
- 2018: Yandex releases Europe’s first robotaxi service in Innopolis
- 2018: Waymo announces first commercial service in Phoenix
If you want to stay informed about latest technology and science, as well as imagine a utopia with us, feel free to join the ride on these UTX channels:
YouTube | Spotify | Twitter | Apple Podcasts | Instagram |All Links
Technology
Now we will take a closer look at the technology behind autonomous cars.
The Different Levels of Driving Automation
- Level 0: The system may issue warnings and may momentarily intervene but has no sustained vehicle control
- Level 1: Driver and automated system share control of the vehicle; includes cruise control, adaptive cruise control, parking assistance, lane keeping assistance and automatic emergency braking are examples of Level 1 Driving Automation
- Level 2: The automated system takes full control of the vehicle (accelerating, braking, and steering), but the driver must monitor the driving and be ready to intervene immediately
- Level 3: The driver can safely turn their attention away from driving tasks; the drivier must be prepared to invtervene within some limited time
- Level 4: Like level 3, but no driver attention required at all; the driver may go to sleep; but the in areas where self-driving is not supported, the vehicle must be able to safely abort the trip if the driver does not retake control
- Level 5: No human intervention is required at all; like a robotic vehicle that works on all kinds of surfaces, all over the world, all year around, in all weather conditions
The Features of Driving Automation
- Navigation
- Directional control: Automated Lane Keeping System ALKS
- (Adaptive) Cruise control
- Collision / obstacle Avoidance System (CAS)
- Emergency braking
- Signage and road work detection/recognition
- Traffic detection and avoidance
- Lane assistance
- Parking assistance
The Software Behind Driving Automation
There are 3 main software concepts crucial for driving automation.
Bayesian (dependent probabilities) SLAM (simultaneous localization and mapping) algorithms mapp a vehicle’s location based on the localization of other moving objects
The Real-Time Location System (RTLS) identifies and tracks the locations of other objects based on radio waves, infrared or ultrasound; this does not include speed, direction or spatial orientation.
Sensor fusion relies on the fact that the information of sensors is probability-based. Thus it combines information from several sensors yields higher-probability of correct information.
Driving automation also relies on an aspect of learning. The computer vision is based on deep learning. Driving automation also relies on reinforcement learning. To learn more about them, click on the respective links.
The Hardware Behind Driving Automation
Autonomous vehicles are high-tech devices and use all or some of the following:
- Long-, medium- and short-range radars
- Motion sensors
- Ultrasound
- Laser light
- Geopositioning
- Artificial Intelligence
- Cameras & Computing vision
- GPS and maps
- Lidar (illuminating object with laser light and measuring distance based on reflection)
- Sonar (sound navigation and ranging to detect objects based on emitting and receiving sound echo)
- Odometry (estimation own position and changes by motion sensor feedback)
- Inerial measurement unit (IMU) (measures angular rate and orientation based on gyroscope)
- Stereo vision/stereopsis (3D vision of the environment)
- LED and light systems to communicate with humans/unconnected car drivers
- Systems of authentication
- Networking capabilities with other cars, infrastructure, or centralized institutions
Applications
The applications of autonomous driving are manifold.
They can be used in private cars, public transport, robo-deliveries, long-distance trucking, and even racing.
Another application are traffic platoon systems. Autonomous vehicles can use their communication networks to avoid collisions and manage traffic congestion. It can also be used to read sensory data of other traffic objects.
Networking effects can also be used for platooning. Platooning can be used to coordinate cars to travel in conveys by utilizing network-to-vehicle, vehicle-to-vehicle and vehicle-to-network connections. This would allow for an automated highway system, with increased capacity and sharing resources, and reduced congestion. Platooning would also lead to a more efficient use of fuel due to the reduced air resistance. With the combination of 5G, autonomous vehicles will have enhanced sharing capabilities.
Market Players
The global autonomous vehicle market size is projected to be valued at $54.23 billion in 2019 and is projected to be $556.67 billion by 2026; thats a CAGR of 39.47%
In Silicon Valley, the biggest autonomous vehicle maker is Tesla, and they are aiming to provide a fully self-driving car.
The FAMGA companies have all invested in mobility services, autonomous driving and connected car tech, with Facebook being the noted exception.
Alphabet, in collaboration with Waymo, is one of the leaders, as it has travelled the most distance and has the most distance between disengagements.
Nvidia empowers autonomous vehicles to process large volumes of sensor data and make real-time driving decisions
Other players in this industry are Uber, Lyft, Aptiv, Motional, Nuro, Cruise (GM), Zoox, AutoX, Aurora, and others.
In China, Baidu, Pony.ai, Tencent, Didi, Nio and Beijing New Energy are the biggest players in this industry.
The largest car manufacturers and suppliers (who will also be relevant) are BMW, Nissan, Ford, Delphi US, Daimler , Mercedes, Audi.
We also have Yandex (Russia), Robotaxi (russia), Hanwha (South Korea, planning to launch a 5-seat drone vehicle by 2023), Navya France (autonomous busses), Bosch, Robocar (the first electric autonomous racing car), Otto, Starsky Robotics and Ocado.
Benefits
The Benefits and challenges depend on regulations, where resources flow, society’s adoption of sharing models, society’s development of demand for mobility and travel in general and how other forms of mobility will develop and compete with AVs.
Safety
Autonomous vehicles would lead to a 90% decrease in road safety incidents. Road accidents are the main mortality issue for young people in developed world. This accumulates to 10 million lives saved every decade! Interestingly, 13% of organs donated in the US come from car crash victims.
Further, autonomous vehicles free up healthcare resources that can be used elsewhere. AVs would also lead to improved safety and navigation in dangerous areas for rescue and emergencies, as well as exploration and military.
AVs could also enable a way of simplified identification of AV passengers for the police and authorities, but this would increase the risk of mass surveillance.
Economic
AVs would also create less traffic congestion, a more efficient use of roads, and a higher capacity for roads (roughly 200 to 500% higher capacity). AVs also lead to increased convenience and economic benefit from being able to use spare time when commuting.
There would also be a reduced cost of transportation systems and competition to e.g. trains. It would also lead to a more efficient use of parking land, as time of use of AVs would be dramatically higher and thus less parking time in comparison to non AV vehicles. The freed up space (which in LA accounts for 14% of all land) could be used for parks and recreational areas, regaining natural habitats.
There is also an increased housing affordability in urban areas, as commutting becomes less of a hassle. There would be an increase in asset efficiency through lifecycle and maintenance prediction. Interestingly enough, the advertisement, media and entertainment industry can use the freed up attention of passengers to redesign the AV as a means of entertainment.
Sustainability (ESG)
AVs would lead to a decreased energy demand due to a 20% to 40% increase in fuel efficiency. AVs would also promote a car sharing economy, leading to a reduced number of cars. (However, increased urban travel and increased mobility could reduce or even reverse the benefits).
Another issue will be the increased demand for lithium batteries and electricity. AVs could also create jobs for workers in low income countries to train autonomous systems. AVs would also lead to an increased mobility for disadvantaged groups such as the young, the elderly, the disabled and the poor.
Challenges
However, autonomous vehicles are not without challenges.
Economic
Beside massive layoffs in the taxi-industry, there will definitely be monopolies and oligopolies amongst the high-tech firms and/or vehicle manufacturers.
Societal
Many believe that autonomous vehicles would lead to a decrease in manual driving knowledge, but this is a question of societal adaptation. Only 25% of people actually trust the parking assistance software even though it performs much better. For instance, the MCAS software (which led to 2 plane crashes) led to a large media scandal, but no one pushed to remove all the sensors from the airplane.
Public perception is also linked to reverse survivorship bias, as every accident becomes a massive media scandal.
AVs would also lead to an increase in unemployment (and possible the disappearance of entire industries) amongst drivers, car manufacturers, car insurers, petroleum workers, crash repair shops, public transport workers, roadside restaurants, short haul flights and car rentals.
Another interesting aspect will be our perception of cars. Cars are a symbol of status and associated with skill and fun experiences. With cars becoming part of the mobility infrastructure, this will definitely change.
Legal
AVs will lead to an uncertainty of future outcome and policy design in terms of urban planning, zoning and pricing for regulators in terms of infrastructure planning for e.g. fully driverless vs. partially driverless scenarios.
There will also be massive changes to road traffic policies and conventions.
It is also thinkable that the legal liabilities of vehicle, network, maps, software, sensors and radars and even road system will be shared and split between car, sensor, and software manufacturers, AV system aggregators, the government, and passengers
Accident liability and insurance issues will also see some changes.
Technological
One technological challenge will be the susceptibility of sensors and naviagtion systems to rainfaill, hail or slow which can impact sensor’s measurement abilities.
The complexity, storage, computing and performance requirements for vehicle architecture caused by volume, (real-time) velocity and (format) variety of information (a big data problem) must also be taken into consideration.
There is also the complexity and performance requirements (e.g. delay) for network architecture based on large amounts of moving objects.
Another concern is the complexity and somewhat randomness of exact decision making.
AVs and network also need to be secured against any hacking attempts and general robustness (avoiding system failures).
They would also need millions of kilometers of testing (assessed based on disengagements, human interventions).
Another aspect is privacy. Mass surveillance and usage of location & mobility data as well as consumption, speech and video monitoring data needs to be handled adequately.
The assembly of an AV necessitates a more complex supply chain for cars and networks, necessitating mechanics to learn new skills.
Also, tax payers will have to finance massive and expensive changes to road infrastructure.
The current inability to translate ethics into software code and make it understandable to AI will also be a challenge.
Ethical
Ethical dilemmas (such as the Trolley problem) between means and result, i.e. doing nothing and killing 3 people (deontologist, conforming to rules, optimizing action) or doing something and killing 1 person (utilitarian, optimizing result) have to be discussed.
The MIT has simulated that in the context of autonomous cars and people’s decision making, the results are similar to dogs being prioritized over criminals over cats.
Another potential paradox is when people promote utilitarian ethics, but not if they are inside the car (tragedy of the commons).
Drivers and passengers
Distinguishing between fully autonomous and partially autonomous self-driving cars (which are autonomous) and driver assistance systems (which are automated) is important, as it might lead to driver confusion, distracted driving and accidents
Lack of awareness in the begining is also critical, as being unprepared in the moment of takeover coupled with de-skilling and lower levels of driver experience could have fatal outcomes.
There might even be a risk compensation: people (in their role as drivers or as other traffic participants) might engage in riskier behavior, as they trust in a safe system to counterbalance their risky behavior.
The Automobile Association and Insurance associations advocate for awareness to increase safety. SAE International classification of 2014 by 6 levels with changing reaction time for human intervention and autonomous control per level.
The Utopian Day In The Life
The Utopian Day in the Life would be waking up in Utopia, the first Smart city.
The delivery of packages now happens through autonomous vehicles and is accurate and near-instant. With AVs, entire supply chains have become autonomous.
Driving has become a thing of the past. Everyone takes either self-driving cars or flying taxis and is able to enjoy entertainment, enjoy the view, or work. Traffic jams are only visible in depictions in the museum. Cars normally travel at speeds of 250 kmh.
Increasingly, cars have become assets. They now become taxis during the day, earning money instead of standing in the parking lot 95% of the time. And because it is possible to work in the car, the working hours are shorter.
Driving cars manually has become a form of casual entertainment. I often spend the evenings watching autonomous racing competitions between different manufacturers, each showcasing their hardware and software.
Stay Utopian, Your Techniax!
