Autonomous Vehicles, Artificial Intelligence and The Internet of Things

We will soon see an autonomous vehicle breakthrough fuelled by the current pace of change happening with artificial intelligence (AI), the Internet of Things (IoT), and other technological innovations.

Three companies are among the leaders in paving the way to this disrupted future. They are Tesla, Waymo, and Zoox (Amazon).

Tesla Brings AI to Autopilot

Tesla’s latest self-driving software contains AI that learns autonomously by studying the behaviour of human drivers. States Dhaval Schroff, a Tesla engineer, “It’s like ChatGPT, but for cars.” Using neural networks, the self-driving software simulates billions of driving scenarios.

Historically, Tesla’s Autopilot used a rules-based approach, a structure where specific situations triggered structured responses. Examples included how to respond to lane markings, pedestrian movements, other vehicles, traffic signals, construction signs, speed limit indicators and more. Each of these was given programmed responses.

The new Autopilot doesn’t follow these codified responses to programmed scenarios. Instead, it has learned by watching real-life examples of human drivers. As a result, Tesla is using a new metric to measure autonomy success. It is “miles driven without human intervention.” 

The goal is to have the Autopilot exceed human driver responses. To achieve this, Tesla used a neural network to help the Autopilot teach itself. It approached autonomy with a blank slate. It had no definition for what is a car, bicycle, pedestrian, or any other driving object. Instead, it learned by watching human drivers deal with billions of real-world situations. It taught itself to read signs. It learned the objects of the road and what drivers encounter. It had a library containing data from nearly 2 million Tesla electric vehicles. This treasure trove of real-world examples was its training wheels. How far away is Tesla from launching this latest iteration of autonomy? Currently, it is testing 5,000 vehicles using the new AI Autopilot.

Waymo Approaches Vehicle Autonomy

Waymo began in 2009 as a Google internal experiment. From the start, its goal was to create robotaxis. It builds into vehicles, cameras, Lidar laser technology, real-time sensor data, and custom maps to help visualize the world. Its onboard systems can process billions of bits of data in real time. Its vehicles have been tested, having travelled over 19.3 million kilometres (12 million miles) in real-world and simulated environments.

In August 2023, Waymo launched driverless robotaxi services in San Francisco. The Waymo One app allows pedestrians to book a ride much the way they would call up an Uber. The San Francisco fleet of 250 outfitted Jaguars features high-tech sensors and cameras. Fares range from $18 to $21 U.S., on par with other ride-hailing services.

In March 2024, Waymo received approval to extend its service to a larger area of San Francisco and to Los Angeles where it is operating 50 vehicles in the heart of downtown and Santa Monica. Currently, Waymo in L.A. is restricted to regular streets. Its vehicles cannot drive to the airport or along the many Los Angeles freeways that crisscross the city. That soon will come.

Zoox And Amazon’s Autonomous Vehicles

When Amazon decided to get into the autonomous vehicle market it bought Zoox in 2020, a six-year-old company focused on making mobility-as-a-service in cities a reality. You can see a picture of a Zoox robotaxi at the beginning of this article and no doubt you noticed that it doesn’t resemble any vehicle you have ever driven. That’s because Zoox is a reinvention of the car.

Zoox vehicles were never intended to have a driver. They are designed for full autonomy in dense, urban environments. Like Waymo they feature cameras, radar, Lidar and onboard sensors. But unlike Waymo they are unconventional in how they operate with features like bi-directional driving and four-wheel steering.

If I were placing bets on what robotaxis will look like as we move from car ownership to mobility-as-a-service, I would put my money on Zoox-type designs dominating the future of the automobile.

Why Robotaxis Matter in a Decarbonizing World

Consider how often you get into your car. On average, we use our vehicles 5% of the time. The rest of the day they stay parked on driveways or in parking lots. Parking lots occupy land that can be repurposed. Donald Shoup, an urban planning professor at UCLA notes that in the United States with its 2 billion parking spots, “the area of parking per car is…larger than the area of housing per human.” 

The Uber of the future is the robotaxi, a mobility-as-a-service that provides point-to-point transportation. It is more personal than a bus, subway or train. As mobility-as-a-service becomes more common, fewer vehicles will drive on urban streets. Once people get used to the efficiency of being chauffeured this way the need for personal vehicles will diminish. That will mean enormous cost savings per household yielding greater disposable family income. It means the house footprint will change with no need for a parking pad or garage (the latter at least not for a car).

In-service feature-rich point-to-point robotaxis will replace multiple vehicles on the roads. With fewer vehicles, it means our rush hours, where parkways look more like parking lots, will become a thing of the past.

For vehicle manufacturers, this massive disruption to their business models will cause many to fail as the ratio of cars per person will decline. That means fewer sales, fewer models, smaller production lines, and fewer jobs in the industry.

Some vehicle manufacturers recognize the coming mobility-as-a-service revolution. General Motors and Audi have considered becoming owners of the cars they build, offering them as robotaxi fleets. Other manufacturers will likely follow.

What are the implications for modern economies built around automobile manufacturing? North America represents the prime example. In 2022, the automotive industry produced 14.8 million vehicles and employed 3.3 million not including those in the supply chain providing materials and services. A robotaxi future won’t need 14.8 million vehicles. Manufacturers won’t need 3.3 million workers.

Mobility-as-a-service through robotaxis will profoundly impact lives as the 21st century unfolds. There are many positives. Our environment will be cleaner. Our air will be more breathable. We will see less of an urban heat island effect in our cities. With fewer cars on the road, we will mitigate climate change. The negatives will be the economic disruptions to manufacturers and workers for which current governments and society appear unprepared.