The brave new world of Autonomous Vehicles
Driverless transportation has the potential to save the world, but just how feasible is it right now?
The buzz has been around: autonomous vehicles will change how we move, live and work.
How valid is this claim?
For a start, consider this: according to the United Nations Framework on Climate Change Convention, the transportation sector was responsible for 27 percent of US greenhouse gas (GHG) emissions in 2010. GHGs are one of the leading causes of the greenhouse effect worldwide.
Fuel sources used for road transportation, such as petrol, diesel, and natural gas, produce several GHGs in byproducts. Gaseous emissions from burning these energy sources, including methane (CH4), carbon dioxide (CO2), and nitrous oxide (N2O), can survive in the planet’s atmosphere for several decades, causing continuous global warming.
In European countries, the transport sector was responsible for 30.5 percent of GHG emissions and 12 percent contribution of GHG emissions from road transport in 2014. Another study in China predicted that the transport sector alone would account for 84.7 percent of GHG emissions.
These are dire projections. And the hype around autonomous vehicles is tremendous. So, how soon can we do away with the driver? That reality is still some way off.
The building blocks are in use already
The basic technology for driverless cars or autonomous vehicles is already available and in use. Take, for instance, the front-crash prevention systems that have been in use in many models for several years now. They warn drivers of an impending obstacle and apply the brakes if they do not react quickly enough. This is the first principle in autonomous transportation – replacing the driver.
Other features we take for granted today, such as self-parking, Mercedes-Benz’s Drive Pilot that can take direct control of steering and speed if the driver chooses to relinquish command, are advancements developed to automate vehicles truly.
Big players, big ambitions
Tesla CEO Elon Musk’s claims about the advanced stage of self-driving technology that will power its electric cars notwithstanding, Tesla’s director of Autopilot software CJ Moore has admitted to the California Department of Motor Vehicles that such claims do not match the ‘engineering reality’, according to Plainsite. Tesla’s cars remain at Level 2 technology – a semi-automated driving system that requires a human driver’s supervision.
For full self-driving capabilities, Tesla’s autonomous models must navigate the following requirements on city streets:
- On Autopilot.
- On Auto Lane Change.
- Summon (that will move the car in and out of a tight space using the mobile app or key).
- Smart Summon (to navigate more complex environments and parking spaces).
- Traffic and Stop Sign Control (Beta).
- Autosteer on city streets.
So far, Tesla is well behind its set date of end-2021 to launch a genuinely self-driving model.
What is an automated vehicle?
There are five levels in autonomous vehicle technology, according to most experts. But TWI, independent research, and technology organization, details six levels, with Level 0 being a car with no control over its operations and the driver taking complete charge of the vehicle.
The six levels of Autonomous Vehicles are:
- Level 0: This has no automation. The driver is entirely in charge of running the vehicle.
- Level 1: Driver assistance includes keeping a car in a lane or managing its speed. Most automated driving solutions available in vehicles today require human intervention. It puts them at level 1.
- Level 2: Partial automation, such as assisted steering and speed control. These capabilities require a fully attentive driver prepared to take control at any moment.
- Level 3: Here, vehicles boast more autonomy and can make some decisions on their own. However, the driver must remain alert and take control of the vehicle if the system cannot drive.
- Level 4: Artificial Intelligence (AI) self-driving software must substitute for the human driver to move to this level of automation. In this category, vehicles will essentially be self-driving, deciding where and when they will drive based on inputs. An example of this level is an automated taxi plying a specific route.
- Level 5: These vehicles will take users anywhere they wish by just entering the address. Passengers can practically sleep until arrival at the destination. The car will take care of every aspect of the journey irrespective of the type of road, terrain, or condition. Most importantly, at levels 4 and 5, the responsibility for the car no longer rests with the driver but the manufacturer.
According to TWI, “This full automation will be enabled by the application of 5G technology, which will allow vehicles to communicate not just with one another, but also with traffic lights, signage and even the roads themselves.”
Level 4 vehicles are showing significant gains. Google’s Waymo robotaxi project is said to be at that stage now, though it is still being tested. Other companies are also making progress with Level 4 vehicles, though they are not commercially available.
Aiming for a perfect 4
However, considering the costs involved, level 5 will remain a pipedream for now. Companies in the run are aiming for a rock-solid level 4 automation, focusing on providing automated service rather than make computers act like humans.
For instance, the US-based self-driving startup Voyage – recently acquired by the General Motors subsidiary Cruise – has released footage of its autonomous vehicles operating for seven continuous hours without a single intervention.
Voyage’s uncomplicated system consists of two units. The ‘Commander’ pilots the vehicle through the streets, while the ‘Shield’ responds to unexpected safety hazards, overriding the Commander in such instances. If the vehicles face a situation beyond their capability, Voyage has a backup in a remote pilot that will operate the vehicle. Voyage focuses on low-speed, well-mapped neighborhoods to ensure that its safety systems function correctly without accidentally endangering lives.
Likewise, Daimler has partnered with Google’s self-driving car initiative Waymo to develop level 4 automated trucks for fleets. Meanwhile, GM recently started testing its autonomous vehicles without safety drivers on designated streets in San Francisco, USA. The car company optimistically plans to have driverless vehicles out on the roads in a year.
Are self-driving cars safe?
So, how safe are these level 4 cars? Waymo claims trials more than 33 million km of autonomous driving as of January 2020. This figure may sound reassuring, but with the current death rate due to road accidents in the US at 1 per 160 million km, experts opine that this is too small a trial to sufficiently demonstrates safety for human drivers.
According to this analysis, the pandemic has been a recent challenge leading to a temporary suspension of real-world testing for data. It can result in significant delays, according to this analysis.
Others believe that deploying relatively safe self-driving cars will save lives by getting rid of human errors.
With 90 percent of the accidents said to occur due to driver error, autonomous vehicles are expected to be safer.
Overcoming the barriers
While the future of Level 4 self-driving systems looks promising, producing an autonomous vehicle that can complete a journey safely and legally without human input has its challenges.
The key barriers to be overcome before AVs can be mass introduced to the market include technology, infrastructure, regulations, and public acceptance.
Technology is the determinant of how successful autonomous vehicles will be. Sensors, integration of Machine Learning, LIDAR, and a host of other technologies are required to put an autonomous vehicle on the road.
The issue, according to experts, is that the industry remains primarily self-regulating, particularly in determining whether the technology is safe enough for use on roads. Policymakers and regulators around the world have been struggling to keep pace, unable so far to determine the criteria for automated vehicles’ technology.
A critical aspect of regulation and policy will be on liability insurance, says this report. “One of the murkiest areas for self-driving vehicles is the issue of liability and insurance. How will insurance companies handle fender benders while a driver is reading and not paying attention to the road? Furthermore, who will constitute as the “driver;” who has ultimate “control”?
Comprehensive safety testing and evaluations are required before manufacturers can test self-driving software in real-world conditions. The onus would be on regulators to devise tests and ensure that companies benchmark their algorithms on standard data sets before their vehicles are allowed on open roads.
Experts advise a graded approach to certification requiring a self-driving system to be first evaluated in simulations and then in controlled, real-world environments before being allowed to operate.
The biggest hurdle to overcome would be public acceptance. For this, the public would have to be involved in all self-driving vehicle deployment and adoption decisions. The public’s trust will be undermined if self-driving technologies are not regulated to ensure their safety.
“Public trust is a crucial issue for self-driving cars. Early on, the public seemed optimistic that robots could outdrive the proverbial “nut behind the wheel.” However, a series of unsettling minor accidents like hitting a stopped firetruck slowly eroded trust and highly publicized fatal crashes scared the public,” says this analysis.
The road ahead
With the expected arrival of autonomous vehicles on roads, governments and urban planners have begun to think about how this will change the way people live and work and affect the cities.
For instance, Chandler city in Arizona, where Waymo is testing its fleet, became one of the first US cities to rewrite its zoning code to facilitate autonomous vehicles. This has influenced real estate, too, with developers now applying to build fewer parking lots. Instead, they will provide passenger loading zones with benches and trees for shade, entailing a considerable saving on their projects.
The advent of autonomous transportation can thus be an opportunity to evolve from an auto-oriented society to a more pedestrian-oriented community, making cities more beautiful by easing up the traffic and eliminating space-consuming parking lots. It will lead to greening urban areas and suburbs as commuting becomes easier and eco-friendly. With fewer cars on the road, governments may be able to reallocate a good portion of the billions spent annually on maintaining highways and roads.
However, this reality is some way off. According to this report, “Although autonomous car companies have made phenomenal technological progress, it will still take many years for fully autonomous cars to become available to the public. Specifying a definite year may not be possible at this stage. According to some predictions, cars may become fully autonomous by 2035.
Top players in the Autonomous Vehicles market (AV market)
Pony.ai
Pony.ai is an autonomous vehicle technology entity co-located in Silicon Valley, USA, and Beijing and Guangzhou, China. A startup that has attracted significant funding, Pony.ai builds complete autonomous driving solutions. It is leading the way in autonomous mobility technologies and services across the US and China, with robotaxi pilots in both markets. Pony.ai is on its way to creating a safe and sustainable mobility future by delivering autonomous mobility across both markets.
Volvo
Volvo is a pioneer in safety technologies and was one of the first car manufacturers to bring innovative safety devices. Lane centering to its complete vehicle portfolio as well as develop the Pilot Assist II advanced autopilot-like functionality.
Volvo, the second-biggest manufacturer of semi-trucks, has been adding partially autonomous features in its trucks but is yet to build fully driverless delivery vehicles.
Volvo is now partnering with self-driving startup Aurora on a new line-up of fully autonomous semi-trucks to be used in North America on hub-to-hub routes.
Aurora
Self-driving startup Aurora, founded by former executives from Google, Tesla, and Uber, has been testing its ‘Aurora Driver’ hardware and software stack in its test fleet of minivans and Class 8 trucks in the US since 2020. Unlike its rivals, which are primarily focused on robotaxi applications, its first commercial service will be in trucking.
Aurora has raised $690 million in funding so far. It has acquired Uber’s autonomous division and is in partnership with Toyota and Denso to develop a fleet of robotaxis, the first of which is expected to begin operations by the year-end.
Co-founder and CEO Chris Urmson helped pioneer Google’s self-driving car initiative, earning him the title ‘Henry Ford of autonomous vehicles. Another co-founder, Sterling Anderson, helped lead Tesla’s Model X project. At the same time, their partner Drew Bagnell ran a research lab at Carnegie Mellon before working on Uber’s autonomous vehicles. Fiat Chrysler, Hyundai, and EV start-up Byton are also Aurora customers.
Argo AI
A relatively new entrant, Argo AI has a growing footprint in the US. It has partnered with Volkswagen and is now testing across several cities.
An independent company launched with a $1 billion investment from Ford in 2017, Argo AI is not trying to develop a car or operate a service. Its goal is to create a self-driving system through a combination of sensors, including LiDAR remote sensing, radars, and cameras.
GM Cruise
With investors like SoftBank, Honda, GM, and T.Rowe Price Associates, Cruise is also a big name in the autonomous vehicle game. It already has a significant number of vehicles on the road, and its self-driving cars are in the latter stages of development. Cruise claims that its self-driving Chevy Bolts are built from the ground up, with 40 percent of its parts altered to facilitate autonomous driving. With GM’s backing, Cruise is also in pole position to roll out vehicles on an assembly line basis once they are approved.
Cruise vehicles use Machine Learning (ML), cloud-based tools, and Internet of Things (IoT) sensors to gather data on their surroundings.
