The revolutionary power of artificial intelligence (AI) continues to expand across industries, including what may drive the future of your daily commute. Autonomous vehicles have emerged over the last several years as a promising look into what is next for the transportation industry, with artificial intelligence playing a prominent role in its development.
Autonomous vehicles incorporate AI through sensors, cameras, lidar and radar to help with navigation. These functions are fulfilled by creating environmental maps and predictive models, all supported and refined by reinforcement learning at different levels, “a type of machine learning technique that enables a self-driving car to explore and interact with the dynamic environment and make decisions based on the data attained by experiences.”
No vehicle currently on the market is fully autonomous. According to Statista, “In 2019, there were some 31 million cars with at least some level of automation in operation worldwide.” There are six official levels of automation that a vehicle can fall into, with many cars on the roads today falling into Levels 0-2. The levels range from Level 0, No Driving Automation, to Level 5, Full Driving Automation (or fully functioning, self-driving vehicles), progressively gaining more and more automation as you move through them. The current levels, 1 through 4, encompass:
Level 1 = Driver Assistance: includes at least one driver support with strong driver supervision
Level 2 = Partial Driving Automation: more advanced-version systems of those in place in Level 1, still requiring driver supervision
Level 3 = Conditional Driving Automation: the car is driven entirely by the software and does not require constant driver supervision, but there does still have to be a driver in the vehicle on standby
Level 4 = High Driving Automation: does not require human drivers to operate the vehicle, even in case of emergencies, although some may require human intervention
Waymo, a leader in the development and deployment of autonomous vehicles, has advanced significantly due to the power of AI innovation in this space. They currently operate Level 4 vehicles in Phoenix, San Francisco and Los Angeles, and they will soon expand to Austin and Atlanta. This is also accompanied by an exclusive partnership with ride-share service Uber, carrying even more Americans into the future of transportation together.
The company recently released a research paper detailing the development of its new training model, EMMA, which is built upon Google’s Gemini. The model can incorporate a greater reasoning capability than previous technologies in autonomous vehicle innovation. This capability relies upon “chain-of-thought reasoning” to mimic logical human decision-making, improving driverless vehicles’ trajectory prediction, object detection and road graph understanding.
Improved reasoning capabilities are just one of the many benefits generative AI in autonomous vehicles presents to American consumers. Further advantages include:
- Improving Road Safety
In the United States, human error accounts for 94% of all crashes. Implementing AI into vehicles would help mitigate this extremely high percentage rate while still basing driving decisions on logical reasoning. A study done last year in San Francisco demonstrated that human drivers had a crash rate of 50.5 crashes per million miles compared to 23 crashes per million miles in self-driving vehicles. Part of this higher rate can be attributed to distracted driving, which currently costs the United States around $165 billion yearly.
- Reducing Traffic
According to UC Berkeley professor Alexandre Bayen, “In a recent study, when an automated car led human-controlled vehicles, stop-and-go traffic was eliminated and gas usage was reduced by 42%.”
- Increasing Accessibility
Many Americans are unable to drive due to disabilities, age, etc., significantly stunting their mobility and overall access. A national poll released in 2018 notes that 1 in 5 Americans older than 65 cannot drive, alongside over 80% of young adults with disabilities who state that “they’re often prevented from doing the activities that they’d like to do because of lack of transportation or inability to drive.” Allowing innovation to continue with AI integration into autonomous vehicles opens the roadways to a significant subset of the population. This includes the blind and visually impaired to ensure adequate accessibility features are available, providing them with a newfound sense of independence many may have never experienced before.
As with all new advancements, associated concerns also need to be addressed. With AI integration in autonomous vehicle development, potential consumers must have trust and confidence in these products.
A recent Johns Hopkins study demonstrates that public perceptions are generally timid when new and different technologies are introduced on the market. Only about 26% of people had initial positive opinions about autonomous vehicles. The study stresses the importance of education and building awareness in cases like this. As noted in their research, “when participants learned about and then considered potential societal benefits, particularly in serving vulnerable populations, support for AVs nearly doubled to 50%.”
Autonomous vehicles are clearly a growing and evolving part of American business and life. AI integration in the field could help significantly reshape transportation from daily commuters to ride-share enthusiasts alike. Cities like Phoenix, Austin, Los Angeles and more are already testing autonomous vehicles on their roadways, with the potential for the rest of the country to follow in the coming years as we all navigate this new transportation landscape and ride into a better future together.