In 1935, David H. Keller wrote a short story that summarized the dream of an autonomous car:
“Old people began to cross the continent in their own cars. Young people found the driverless car admirable for petting. The blind for the first time were safe. Parents found they could more safely send their children to school in the new car than in the old cars with a chauffeur.”
— The Living Machine (1935) by David H. Keller
The dream, as we know it, is not far from our current reality. For instance, Volvo introduced its full autonomous test vehicle in 2017 and plans to bring its unsupervised autonomous vehicle (AV) by 2021. Google’s Waymo, which started its journey in 2009, has completed 20 million driving miles. In 2014, Tesla announced that its car would be capable of self-driving about 90% of the time. Today, all Tesla models are equipped with self-driving capability.
The dream lingers on, but not everyone believes in it. An article from Wired in 2017 titled “No one knows what a self-driving car is, and it is becoming a problem,” and another one from the Verge in March 2020 indicates that “Americans still don’t trust self-driving car.”
There is a gap between the dream and the public understanding. This gap is caused by the auto industry, which does a terrible job convincing the public how the new systems work.
The best answer to this problem came from Amazon Web Services (AWS) in 2018. AWS introduced a 1/18th-scale vehicle called DeepRacer. This miniature car comes with a stereo camera, a dual-core Intel processor, and a Lidar sensor, making it capable of solving various tasks.
The Deepracer is designed to make it easier for programmers to get started with reinforcement learning. Reinforcement-learning is a set of algorithms that pick up skills through trial and error using feedback from its actions and experiences.
Matt Wood, an executive at AWS, hopes DeepRacer will help coders get a feel for reinforcement learning. He wishes that programmers would then apply reinforcement learning to industrial cases, such as wind turbine optimization or container scheduling in ports.
To play with DeepRacer, we must first train the code in a virtual world created by Amazon. We must write algorithms that reward the vehicle when they do something right, such as winning a race or avoiding an obstacle.
This year, Ray Goh from Singapore won the F1 ProAM Deepracer tournament, which featured Daniel Riccardo, the F1 racer. There is an inspiring video titled “Racing to a dream” from Ray about his journey of winning the competition.
I do not know what “dream” in that title means, but it somehow reminds me of David Keller’s words about autonomous cars from 1935. Somehow his dream remains. With our technological advancement and the Deepracer growing popularity, I hope many more people would share the dream of autonomous cars.