How long will the transition to autonomous vehicles take?
I recently attended the first AutoSens Conference held in the United States. AutoSens not only delivered a quality conference, but also a compelling experience. Detroit was selected to host this event as the historic pioneer of the last revolution in mobility -- mass production of the automobile via the assembly line. Also, Detroit is now leading the way in the next renaissance -- that of a mass-produced autonomous vehicle.
One of the highlights of the conference was the venue in the form of the M1 Concourse race track, which is essentially a country club for auto aficionados, complete with private garages and a state-of-the-art 1.5-mile performance track. The conference also included autonomous test drives provided by Dataspeed, a company who specializes in driverless vehicles.
To further build upon the experience, a dinner event was held in the Henry Ford Greenfield Village, a historical tribute to great inventors including Ford, Edison, and the Wright brothers, to name a few. On display and sharing the road with the Dataspeed self-driving car was an authentic, functional Ford Model T, the first assembly line automotive vehicle, which revolutionized cars and made them the primary form of personal transportation today. Seeing this exquisite car side by side with autonomous vehicles got me pondering the similarities and challenges that must have occurred with the transition from horse-driven carriages to cars, and the transition from human-driven cars to autonomous vehicles.
The juxtaposition of a Ford Model T (circa 1917) with pair of 2017 self-driving cars powered by Dataspeed (Source: OpenBoxPhoto.com, courtesy of Dataspeed)
The earth-shattering revolution of the automobile
A century ago, when normal carriages were pulled by horses, the concept of a car without a horse to pull it was perceived as a radical novelty. Like most revolutionary inventions, the automobile was not accepted with open arms. It was considered dangerous, imposing, and a nuisance to the horses, which were the main mode of transportation. The path to cars taking over the roads, replacing horses, buggies, carriages, and wagons was neither swift nor simple. This hilarious video depicts a law that required motor-vehicle drivers to dismount at every intersection to announce their presence by making loud noises.
The law dictated how this should be executed, including discharging a firearm or other form of explosive, among other interesting methods. This would pose serious problems in rush-hour traffic today!
Prudent investors were also averse to the automobile venture. According to this article from AmericanAutoHistory.com, in 1903, the president of the Michigan Savings Bank advised against investing in the Ford Motor Co., based on the view that "The horse is here to stay but the automobile is only a novelty -- a fad." I think we'd all hope for better insight from our financial advisors.
"The horse is here to stay but the automobile is only a novelty -- a fad." -- The president of the Michigan Savings Bank, 1903 (Source: Unsplash.com)
From today's perspective, a century in the future, it's easy to scoff at this skepticism. However, the reluctance to embrace motor vehicles then is not very different from the hesitance towards autonomous vehicles today. In the future, the current guarded and sluggish advances towards regulating autonomous vehicles may be viewed as excessive over-cautiousness.
Eventually, as you've probably noticed, the car did dominate, as evidenced by the absence of horses on the roads, and I expect autonomous vehicles to follow a similar path. Using the timeline of the motor vehicle versus the horse as a guideline, a gradual increase of self-driving cars will soon lead to them becoming the primary form of personal transportation, especially in cities and highly populated areas. Further, we should expect to see manually driven cars eradicated within about thirty years, except for hobby and leisure, as was the case with horses. As my colleague, Gunn, amusingly describes, even car racing may be taken over by autonomous vehicles.
Self-driving cars are the next big change; mass production is key
Automobiles had been around for quite a while before the Model T, but were only obtainable by the upper class. Though many rich individuals owned cars, they couldn't do much with them due to lack of infrastructure and confining laws. Once the assembly line was perfected, and the number of cars grew significantly, there was no choice but to pave roads and regulate traffic accordingly.
To instigate the next revolution, there is still a major step ahead. Self-driving cars are making a lot of headlines, but they are not yet threatening to replace all the vehicles on the road. For that to happen, the autonomous vehicles will benefit from mass-production and competitive pricing. As cost improves for self-driving cars, so does the availability to utilize self-driving transportation for more people.
Can you trust artificial intelligence to drive you around?
If I continue with my analogy, skepticism towards the automobile a century ago is today complemented by the fact that autonomous vehicles will be powered by another technology of which people are also skeptical -- artificial intelligence (AI). In addition to the preconceived idea that robots will one day take over the world, car manufacturers must also overcome the natural human instinct not to trust autonomous technology. What if it's faulty? What if it breaks down? What if it doesn't react in time? These are all legitimate concerns, both regarding trusted steeds versus the first automobiles, and trusted cars versus autonomous prototypes. Today, it is clear to anyone that a motor vehicle is safer and more effective than a horse. It took a while to get there and required roads, traffic lights, and signs, but now people are so comfortable in their cars that the problem becomes staying focused on the task of driving.
So, what about AI? Deep learning and other forms of machine intelligence are becoming ubiquitous in mobile and embedded devices, as well as voice interfaces, and many other aspects of our lives. These technologies are constantly improving and are surpassing human achievements in an increasing number of fields -- from the game of Go to image recognition -- and they will certainly lead to safe autonomous vehicles.
The assembly line made car manufacturing efficient; DSPs are making AI efficient
One of the main challenges of the autonomous vehicle revolution will be to efficiently pack these super-human brains in embedded systems, as opposed to them requiring rooms full of servers with overheated CPUs and GPUs. Size, cost, and power consumption must all be scrutinized to create an efficient and effective solution to make production costs feasible for mass-consumption.
An efficient embedded system based on a vision and imaging processor, such as the CEVA-XM6, for example, can achieve both ultra-low-power and precise computer vision. It's also possible to streamline development using the CDNN deep learning toolkit. You can find out more about how to unleash the potential of these platforms by viewing our on-demand webinar: Challenges of Vision Based Autonomous Driving & Facilitation of An Embedded Neural Network Platform.