Autonomous driving

The concept of autonomous vehicles dates back to the 1920s, with early experiments using radio-controlled cars. The first truly self-sufficient autonomous vehicle, the Stanford Cart, was developed in the 1960s, and by the 1980s, Carnegie Mellon University's Navlab and the EU's Prometheus Project advanced the field. The DARPA Grand Challenges in 2004 and 2005 spurred modern development, with Stanford's Stanley winning the 2005 race. Since then, companies like Google (now Waymo), Tesla, and Uber have invested heavily, and by the 2020s, robotaxi services have launched in several cities. The history is marked by incremental progress in sensors, computing, and AI, with each milestone building on previous breakthroughs.

Autonomous vehicles rely on a suite of sensors, including LiDAR, radar, cameras, and ultrasonic sensors, to perceive their surroundings. These sensors feed data to powerful onboard computers that use machine learning algorithms to detect objects, predict their behavior, and plan safe trajectories. High-definition maps and GPS provide localization, while vehicle-to-everything (V2X) communication can enhance awareness. The system architecture typically includes perception, prediction, planning, and control modules. Redundancy is critical: multiple sensors and fail-safe mechanisms ensure safety. Edge cases, such as unusual weather or unpredictable pedestrians, remain challenging, and the technology is often tested in simulation before real-world deployment.

Safety is the primary motivation for autonomous driving, as human error accounts for the vast majority of crashes. However, ensuring the safety of autonomous systems is complex, requiring rigorous testing and validation. Regulatory frameworks are evolving, with the U.S. Department of Transportation issuing guidelines and the United Nations adopting regulations for automated driving. Ethical dilemmas, such as the trolley problem, raise questions about how vehicles should prioritize lives in unavoidable accidents. Liability, cybersecurity, and privacy are additional concerns. Public acceptance and trust are also critical, and incidents involving autonomous vehicles have led to increased scrutiny. Despite these challenges, proponents argue that autonomous driving could dramatically reduce road deaths and improve mobility for the elderly and disabled.

Beyond the headlines, autonomous driving has many niche dimensions. For instance, the first autonomous vehicle to drive on public roads was a Mercedes-Benz van in 1986, part of the EUREKA Prometheus Project. The term 'robotaxi' was popularized by Google's self-driving car project, which began in 2009. In 2016, a self-driving truck completed a beer delivery in Colorado, marking a milestone for freight. Autonomous racing, such as the Roborace series, pushes the limits of AI at high speeds. In agriculture, autonomous tractors are already in commercial use. The technology also has military applications, with the U.S. military testing autonomous convoys. Moreover, the development of LiDAR has been significantly advanced by the autonomous driving industry, leading to cost reductions and new applications in other fields.

LiDAR

Light Detection and Ranging; a remote sensing method that uses laser light to measure distances and create 3D maps of the environment.

SAE

SAE International, formerly the Society of Automotive Engineers; a professional organization that defines the levels of driving automation.

V2X

Vehicle-to-everything; communication between a vehicle and any entity that may affect it, including other vehicles, infrastructure, and pedestrians.