AV Development Continues, Level 5 Autonomy Coming Soon to Consumer Cars

In 2002, the U.S. government’s Defense Advanced Research Projects Agency (DARPA) announced the DARPA Grand Challenge,^[1] the first long-distance competition for driverless cars in the world. In March 2004, the first race took place in the Mojave Desert, and not a single vehicle could cross the finish line. Most of the cars were disabled or ran into obstacles within the first 20 miles of the race.

The Grand Challenge second race^[2] took place in October 2005. This time, five vehicles successfully completed the 132-mile course. The winner was “Stanley,” a car from the Stanford University Racing Team. At the time, Stanley took almost seven hours to complete the 132-mile distance, at an average speed of 19 mph.

In October 2021, another autonomous car race will occur, this time in the Indianapolis Motor Speedway: the first INDY Autonomous Challenge.^[3] This time, the competition will take place on a race track, and the vehicles are expected to clock speeds over 200 mph. Currently, there are 24 teams from 11 countries participating in the challenge, including the “MIT Driverless”^[4] team, sponsored by Arrow.

While autonomous vehicles have been around for more than a decade, large-scale adoption is still many years away

Over the past 15 years, there has been a continuous and substantial investment in autonomous vehicle (AV) technology. Companies such as Uber, Waymo, and Tesla have developed autonomous cars capable of Level 5 (full driverless autonomy) and demonstrated the potential and safety of AVs. In 2016, Google’s Project Chauffeur, now Waymo, already accumulated nearly 2 million miles of autonomous driving in the United States, testing the technology on city streets, country roads, and interstate highways.

Consumers, however, are not convinced that full driverless modes of transportation are safe to use. A few accidents at the early stage of road testing, including a pedestrian death, cast doubts about autonomous cars’ overall safety.

Meanwhile, AVs have entered service as robotaxis or driverless buses in many cities worldwide. In 2016, MIT spinoff Nutonomy^[5] started testing a driverless taxi service in Singapore, and in 2018, Waymo One,^[6] the first fully self-driving vehicle taxi service, began in Phoenix, Arizona.

Nowadays, organizations developing and producing AVs and advanced driver-assistance systems (ADAS) agree that mass adoption will arrive by combining three factors: innovation, education, and regulation.

Recently, during an interview on CES 2021, Mobileye CEO Amnon Shashua argued that the industry needs “to convince society that these machines make decisions that align with human judgment.”^[7]

Most automakers see AVs as the future and now concentrate on comfort and safety-related ADAS

One of the most disruptive innovations that automakers have been incorporating into their vehicles is ADAS. This new software-based solution goes from information features such as lane-change warning, traffic and hazard data, and route planning to semi-autonomous driving actions such as self-parking, hand-free cruising, automatic braking, and obstacle avoidance.

One example of limited autonomy is GM’s Super Cruise.^[8] The Level 2 partial automation, featured on the Cadillac line and now coming to some of GM’s electric cars, uses real-time cameras, sensors, and GPS, along with LiDAR precision-map data, to enable autonomous driving on the highway. The system is designed to detect every curve, helping the vehicle stay centered in the lane. It doesn’t respond to traffic signals, and the driver needs to pay attention to the road at all times.

Over the last few years, Tesla has been the one company pushing the envelope of autonomous driving. While it is still required for the driver to be able to take control of the vehicle at any time, Tesla’s cars can operate autonomously in most situations, especially in highway traffic. The automaker is aiming to offer full Level 5 autonomous driving within months.

In fact, last July, Elon Musk, during a video interview for the World AI Conference in Shanghai, said that Tesla is really close to offering fully autonomous driving. He told viewers: “I’m extremely confident that Level 5 or essentially complete autonomy will happen, and I think it will happen very quickly. … I remain confident that we will have the basic functionality for Level 5 autonomy complete this year. … I’m absolutely confident that this can be accomplished with the hardware that is in a Tesla today. And simply by making software improvements, we can achieve Level 5 autonomy.”^[9]

Advanced sensors, V2X technology, AI, and massive computer power are the keys to full autonomy

The teams that participated in the first DARPA Grand Challenge had to improvise most of the equipment, including some modifications to the vehicles to control essential functions like acceleration, braking, and steering. Additionally, they featured laser range finders (early versions of LiDAR), radar systems, stereo cameras, and monocular vision systems. Stanley, the second Grand Challenge winner, used seven on-board Pentium M computers to process information and drive itself. Nowadays, we carry 10× the same computing power in our pockets.

Obviously, it is one thing to drive an autonomous vehicle in the desert, with limited obstacles and nobody around. It is entirely different to enable self-driving cars in heavy traffic and on city roads.

In the past few years, another important milestone in the road to autonomous driving has been the arrival of machine-learning algorithms and image processing. AVs now incorporate powerful CPUs, GPUs, and Tensor processors to evaluate all the information from different sources, including 360º cameras and infrared sensors. On-board computers take only a small amount of space under the dashboard and handle all the information to make critical decisions in milliseconds.

The arrival of the latest 5G networks allows for seamless connectivity between vehicles and cars talking to the infrastructure (V2X). Compared with previous generations of cellular connectivity, 5G offers critical functions such as ultra-low latency, signal reliability, and massive IoT.

Additionally, short-range communication using DSRC^[10] standards allows vehicles to talk to each other (V2V) without using cellular towers, enabling additional features such as truck platooning and lane-change coordination.

Technology, government regulations, and people’s trust could help AVs to soon cruise on our roads

Experts agree that the autonomous vehicles on the road will be the ones dedicated to transportation services, such as robotaxis and driverless buses, for the next couple of years.

While several automakers offer some self-driving features and probably Level 5 autonomy soon, there is a lot of work to be done for the personal use of an AV without a driver at the wheel.

That will happen later this decade through innovation, education, and regulation. Manufacturers, telecommunications companies, internet providers, and governments need to collaborate to create a combined ecosystem able to convince people that driverless cars are safer, reliable, sustainable, and life-saving machines.