Driverless | Hod Lipson

Summary of: Driverless: Intelligent Cars and the Road Ahead (MIT Press)
By: Hod Lipson


Welcome to the future of transportation! In the book ‘Driverless: Intelligent Cars and the Road Ahead’, Hod Lipson explores the imminent driverless revolution fueled by advancements in AI and robotics. The book highlights both the benefits and challenges that accompany autonomous vehicles, from life-saving potential and reduced pollution to job losses and privacy concerns. This summary provides a comprehensive overview of how vehicles will evolve from mere transportation machines into smart, autonomous robots, and the impact they’ll leave on our society, environment, and lifestyle.

The Driverless Revolution

In the mid-21st century, a “driverless” revolution will transform the transportation industry. Advances in deep-learning AI and robotics will result in autonomous machines with superior driving awareness and perception. The benefits of increased safety, decongested cities, improved air quality, enhanced mobility, and free time will be contrasted with lost transportation jobs, privacy concerns, and ethical dilemmas. The adoption of driverless cars, which are expected to be “twice as safe” as human-driven cars, will save many lives and mitigate pollution. The challenge for driverless-car engineers is not automating repetitive aspects of driving but engineering for the unexpected and rare cases. As sensing technologies improve, societies will adopt driverless cars gradually and safely.

The Pros and Cons of Driverless Cars

Driverless cars promise convenience, increased usage, and reduced need for parking lots. However, they also bring potential negative impacts such as increased congestion, pollution, and vehicle degradation. Despite major tech giants and auto brands investing in car autonomy R&D, the automotive industry lacks experience in “robotic operating systems,” seeking a gradual shift into driver-assist and limited autonomy. The US Department of Transportation favors a gradual approach but this can lead to complacency and confusion. Nonetheless, federal transportation officials have the potential to throw their support behind driverless cars and save tens of thousands of lives.

The Journey of an Artificially Intelligent Operating System in a Driverless Car

Driverless cars are the future of transportation, and the key to their success lies in the reliability of their artificially intelligent operating system (OS). These systems must be able to navigate and plan ahead in real-time to avoid fatal accidents. Carnegie Mellon University’s National Robotics Engineering Center is one of the many universities working towards solving the challenges that come with the development of such an OS. Researchers are even working to overcome “Moravec’s paradox,” which makes it more challenging to teach “perception and mobility” to a machine than achieving difficult tasks like playing chess. Driverless cars rely on “high-level” controls for planning and navigation and “low-level” feedback for regulating systems like antilock brakes. While “symbolic AI” can handle search tasks, artificial perception requires machine learning to accumulate flexible patterns from vast data sets. Achieving 99.9999% accuracy, which is necessary in driverless cars, is a significant challenge. To ensure humans’ safety, these OSs require redundancy and reliability, similar to avionics systems. The driverless car’s middle-level autonomous controls generate a dynamic “occupancy grid” that assigns uncertainty cones to unpredictable objects like pedestrians, avoiding any unplanned downtime. A “humansafe” standard could be set by federal regulators wherein a driverless car that travels twice the distance of an average human driver would be advertised as having a humansafe rating of 2.0. In conclusion, driverless cars will revolutionize our mode of commuting, living, and shopping, but the success of this technology lies in the OS’s reliability and accuracy.

The Rise and Fall of Electronic Highways

The 1956 Federal Highway Act allowed for the development of the interstate highway system, triggering America’s “car culture.” General Motors and RCA teamed up to create an “electronic highway” that never took off due to high costs and the rise of consumer-safety concerns. Now, with the development of autonomous vehicles, the US DOT is still thinking in outdated paradigms. Instead of costly vehicle-to-vehicle systems, autonomous technology only requires basic infrastructure like clear road markings and cheap IoT technology. Federal regulations could ensure safety, liability, and privacy for this new era of transportation. Several states have already implemented autonomous vehicle driver’s license programs.

Want to read the full book summary?

Leave a Reply

Your email address will not be published. Required fields are marked *

Fill out this field
Fill out this field
Please enter a valid email address.
You need to agree with the terms to proceed