The earliest of automobiles when used to run down the streets were referred to as “horseless carriages”. By the time the term ‘car’ was coined, the industry had already evolved so remarkably, it was hard to believe that this was thought of something impossible just a few decades ago. This cycle has restarted. The idea of being chauffeured around by a string of zeros and ones was ludicrous to pretty much everybody just a few years ago until In January 2014, Induct Technology’s Navia shuttle became the first self-driving vehicle to be available for commercial sale. The following years saw exponential developments starting from the competition for autonomous vehicles put on by Darpa, Google’s first ever self-driving car project Waymo to Tesla’s recent revolutionary suite of advanced driver-assistance system (ADAS) Autopilot mode.
The Technology Behind
Self-driving vehicles incorporate an extensive range of technologies like radar, cameras, ultrasound, and radio antennas to navigate safely on roads. In modern autonomous vehicles, these technologies are used in sync with one another, as each one provides a layer of autonomy that helps to make the entire system more reliable and robust. The subset of machine learning in AI, Deep learning mimics neuron activity which supports functions like voice and speech recognition, voice search, image recognition and processing, motion detection, and data analysis. Working in conjunction, these functions help the vehicles recognize pedestrian traffic, other vehicles on the road, traffic signals and adhere to mapped-out routes. Sensor systems are rapidly evolving to meet the demands of expanded autonomous-vehicle operations, including radar, LIDAR, and cameras. For example, Tesla’s driverless car technology, known as ‘Autopilot’ uses eight cameras to provide 360-degree visibility, while twelve ultrasonic sensors and a front-facing radar work to analyze the vehicle’s surroundings for potential hazards.
Rapid and consistent connectivity between autonomous vehicles and outside sources such as cloud infrastructure ensures signals get to and from the vehicles more quickly. The emergence of 5G wireless technology which promises high-speed connections and data downloads, is expected to improve connectivity to these vehicles, enabling a wide range of services, from video conferencing and real-time participation in gaming to health care capabilities. At peak throughput, 5G promises to be close to 1000% faster than 4G LTE, which will make connection woes such as high latency and long response times a thing of the past. Equipped on autonomous cars, 5G networks will allow for seamless communication from one car to another.
Vehicle to vehicle (V2V)
Latency-free high-speed network will allow autonomous cars to communicate with one another. This will allow autonomous cars to exchange information about their current position, route and hazards on the road. For example, if two cars are traveling on a single lane highway and the car in front, through its onboard sensors, detects a hazardous road condition, that information can be relayed to the car behind so that it can begin braking and adjusting its route. Additionally, with a whole network of interconnected vehicles, traffic congestion could be alleviated since vehicles will be able to make intelligent decisions about their current route to maintain a steady rate of vehicle flow.
Vehicle to Infrastructure (V2I)
Besides communicating with other vehicles, self-driving cars connected to a 5G network will also be able to communicate with different infrastructure elements that make up our roads and other transportation systems. As a simple example, considering parking: while an autopilot car may be able to get you from point A to point B, how will the vehicle know where to park?
With cars constantly coming and going, it’s imperative that self-driving vehicles are able to plan their route in advance therefore information about available parking spaces can be transmitted over the air to a self-driving vehicle through sensors that monitor whether a parking spot is occupied. Once this information is received by the vehicle, that space can then be reserved for that specific car, and this reservation can be broadcasted over the cloud so that multiple driverless cars aren’t fighting for the same parking space.
Vehicle to Pedestrian (V2P)
While communication between vehicles and infrastructure is important, it’s even more important that the vehicles are acutely aware of pedestrians and their exact location. Most of us don’t leave home without our smartphones or some sort of internet ready device (e.g., smartwatch, tablet, e-reader, etc.), which means that in one form or another we are almost always connected to the internet. Interestingly, many of these devices, like smartphones, have the ability to use GPS to determine the exact location of a person. With a 5G network, this information can be instantly relayed to an autonomous vehicle traveling nearby, making it aware of the pedestrian’s whereabouts at all times. With this level of connectivity, driverless vehicles will be able to react dynamically to the position of a pedestrian with collision prevention measures like braking and automatic steering, which in turn should make our streets much safer to travel by foot.
Autonomous technology isn’t a game of zero-sum. There’s an interesting spectrum of five increasingly sophisticated autonomous levels defined below as defined by the SAE International which gives us a brief idea of how far we have reached on the road to fully autonomous cars.
Level 0 – Has no autonomous features. The driver is fully responsible for all operating tasks. The majority of the cars currently on the road are at this level.
Level 1 – This type of vehicle is able to do only one task autonomously, such as automatic braking, lane-keeping, or adaptive cruise control. Drivers are still expected to be fully alert behind the wheel.
Level 2 – Level 2 vehicles are capable of handling more than one task at a time, such as automatic lane-keeping and breaking or steering and acceleration. These vehicles are not considered to have true self-driving abilities and still need human intervention. Level 2 systems currently on the market include the Tesla Autopilot, Cadillac Super Cruise, Mercedes-Benz Drive Pilot, and Volvo Pilot Assist.
Level 3 – These vehicles are able to drive from point A to point B if certain conditions are met. In the case of an emergency, drivers are expected to take control of the car. The only vehicle on the market with level 3 autonomous technology presently available to consumers is the Audi R8, although other automakers are working to develop this type of vehicle for release in 2020.
Level 4 – These vehicles are almost completely autonomous and do not require human interaction. They are constricted by location, cannot surpass certain speeds, and cannot drive in inclement weather. Therefore, a driver or remote operator is still required to be prepared to take the wheel. However, the vehicle is capable of completing a trip with little to no driver interaction making it ideal for fixed route vehicles such as corporate campus shuttles. No cars are currently on the market at this level for consumer purchase.
Level 5 – This is a fully self-driving car that can drive from point A to point B regardless of whether condition or speed. These vehicles do not need a driver, which allows passengers the freedom to focus on other activities such as reading or watching television. No vehicles of this level are currently available.
Autonomous Companies to Watch for in 2020
Waymo
Currently, Waymo operates a Waymo One fleet of Society of Automotive Engineers (SAE) Autonomy Level 4 robo-taxis in Phoenix, Arizona. Prototype testing of fully driverless SAE Autonomy Level 5 vehicles is also taking place concurrently. Waymo has now set up additional testing locations in Michigan and California to acclimatize its vehicles to winter and other conditions.
GM Cruise
With 180 vehicles undergoing testing, General Motors’ Cruise division has the world’s second-largest autonomous fleet. They’ve driven more than 1.6 million kilometers (1 million miles) so far. Although most of the business’ vehicles look like standard Chevrolet Bolt hatchbacks, under the skin, 40 percent of these cars’ parts have been modified for self-driving, according to GM Cruise Vice President Mo ElShenawy. General Motors’ 111-year heritage and experience in car manufacturing gives it a strong edge over dozens of other players.
Argo AI
Ford Motor Company’s Argo AI startup has 100 vehicles undergoing testing in at least six cities in the United States so far. The business has also attracted a $2.6 billion investment from the world’s largest automaker, Volkswagen. Unlike Alphabet’s Waymo or GM’s Cruise, however, Ford and VW don’t foresee Argo AI producing its own vehicles; instead, they want Argo to manufacture self-driving technology for other companies, initially for fleet-based services such as robo-taxis and delivery firms.
Tesla
Tesla has more production vehicles capable of advanced levels of autonomy actually on the road than any other manufacturer. Tesla uses ultrasonic, radar, and 2D camera devices to enable its cars’ autonomous operations instead of LIDARs which are comparatively more expensive. Tesla has more than 600,000 Tesla production vehicles on global roads, having collectively driven more than 3.2 billion kilometers (2 billion miles). Furthermore, in 2019, Elon Musk boldly promised that by the end of 2020, there would be one million fully autonomous (SAE Level 5) Teslas on the road as part of an individual owner/ridesharing network but looking at the current scenario, the plan of action might have to wait.
Baidu
China’s Baidu has more than 300 autonomous test vehicles on Chinese roads having driven more than 3 million kilometers in 23 cities, China’s Baidu like Google, started in the search engine business but soon branched out into other industries. Baidu, has teamed up with Chinese carmaker FAW Group, has inked further agreements with Chinese auto manufacturers BAIC, King Long Motor Group, JAC Motor, and NIO for various partnerships. Similar to Waymo’s service, these vehicles will operate at SAE Level 4, having a driver present but not active within each car. Baidu has said that fully driverless Level 5 operability will come by 2025.
The real question that stands is when will autonomous vehicle technology be ready? The hardware, to start with is mostly accomplished. Radars are cheap and alternatives are already in the talks which are robust enough to build into mass-market cars. Same goes for cameras, and the artificial intelligence that turns their 2D images into computer understandable language. Laser-shooting LIDAR is still pricey, but dozens of startups and major companies are racing to bring its cost to heel. Some have even figured out how to use their photons to detect the speed of the things around them, a potentially key capability. Chipmakers like Intel, NVidia, and Qualcomm are pushing down power requirements for these rolling supercomputers, while companies like Tesla are making their own chips. Autonomous vehicles will bring to market all sorts of new and exciting applications for a variety of industries, like shipping, transportation, and emergency transportation.
By Lavanya Singh, Vellore Institute of Technology, Chennai
