Why those three?
One early indicator of why some technologies have transformative impact on our lives is the fact that they are failing, yet still being developed. It happened many times in the past. Consider all the efforts to conduct manned flight, with all the accidents that accompanied them until that one transformative day in December, 1903. Even afterwards, air accidents continued to happen, and they continue today, but nobody would consider abandoning manned flight.
Early attempts at electric cars produced tiny cars with only 10% of the range of gasoline cars. They were expensive, and required hours of charging. Companies such as "A Better Life," who wanted to build car battery replacement infrastructure failed. Samsung's Galaxy S7 was taken off the market due to exploding batteries. The electric motor, battery, and power control electronics are evolving at a rapid rate, and like any other disruptive technology--we learn from mistakes and do better.
Computer vision is a very broad field, covering everything from your camera's ability to know if your eyes are open or shut before taking a picture, to guiding a rifle bullet to the target, and autonomous cars. This technology heavily relies on processing power and storage capacity (as well as their affordability). As a friend of mine and successful consultant, Tim Durkin, told me: the amount of processing power used to perform a Google search of "how many people walked on the moon?" is higher than the amount of processing power used (and available) to put the first man there... With Moore's law showing no signs of weakness since its inception in 1965, every year, computer vision can do things that could not be imagined the year before. The crash of an autonomous Tesla car that couldn't "see" a white truck turning in front of it was merely a setback, and not a show-stopper.
The enabling infrastructure for cloud computing is really bandwidth. Today, I can communicate with my local hard drive faster than I could with my remote, cloud-based content. However, Internet access becomes faster and cheaper, as optical infrastructure is deployed. More software moves to the SaaS (Software as a Service) model, turning the computer more into a terminal. Adobe completely abandoned the "shrink wrapped" business model for its top-of-the-line software (now called "Creative Cloud") in favor of the SaaS model. My subscription now assures that whenever Adobe releases an update--I get it. Siri does not reside in your phone. It resides in Apple's cloud center in Silicon Valley. Your voice is transferred over the Internet to that center, where much more powerful computers than in your phone can understand you much better that your hands-free car systems.
Where are the opportunities?
With continuous improvements in electric power technology (mainly motor, battery, and power control), I expect that by 2020 electric cars will have very similar performance (including range), if not better than gasoline cars. I expect that to spill over to airplanes (initially private, and then commercial), and ships. Finally, the use of exoskeletons for rehabilitation, commercial, and military purposes should grow as well.
I expect computer vision to "see" better, and be trusted more and more. The FAA has finally allowed UAVs to be used commercially, but only within "line of sight," preventing many applications such as fast package delivery, which Amazon has been touting for a while. The main concern is the interference with commercial aviation and the potential devastating consequence of mid-air collisions. The more reliable computer vision becomes, the more viable autonomous, unmanned aerial vehicles will be. Autonomous cars will not only be the norm, but they will penetrate commercial vehicles (trucks, busses, Uber, and more).
Computer vision would also be deployed through smart cameras to predict terror attacks based on unusual behaviors. The Minority Report is coming.
Internet access started in the 1990s with dial-up rates of 2,400 bits-per-second. Downloading a 10 MB picture would have taken 9.3 hours. That wasn't an issue then because digital cameras didn't exist yet, and hard drives only had 5 MB capacity. Rates of 100 MBPS, 42,000 times faster than 2,400 BPS) are now available as the basic Internet package. In fact, when that 5 MB hard disk drive had cost $500, would you imagine that Google will offer you 15 GB capacity (3,000 more) of Google Drive space for free?
Although video consumption transitions to 4K (and beyond), a 1080P uncompressed stream requires only 1.5 GPBS of connectivity. Compress it, and you can get high quality video stream at less than the basic internet connectivity. Doesn't that mean that we close to where Adobe Premiere Pro software will actually completely reside in the cloud, process in the cloud (like Siri), and my computer would only be the "dumb terminal" that shows the results?
What must happen first?
The first thing that must happen is that smart R&D must be conducted. I didn't use the word "smart" by chance. This is not about money, nor is it about giving employees a day a week to be creative, building an innovation lab, or "driving" innovation. We must motivate creativity, and then allow it. The biggest factor in "smart R&D" is the willingness to try and fail, without asking permission. The second biggest is allowing the employees who know the technology and the market make decisions.
The second thing is an investment in infrastructure. Unlike R&D, this is where billions (if not trillions) of dollars must be invested. For the most part, this is done through companies that raise billions to build infrastructure (and in most cases, such as the early days of broadband Internet connectivity, fail, albeit leaving us with affordable infrastructure), or when the government decides to build and subsidize infrastructure. I hope the President Elect is listening...
Still in the hands of the government, regulations must catch up to technology. The FAA's surprising allowance of the commercial use of drones enabled an industry. Catching up with autonomous technologies and allowing autonomous flight, for example, will enable it further.
To adopt the "infiltration" of those technologies into our daily lives, we need to allow them, trust them, and grow to expect them. Our ability to absorb high rates of technology growth is limited. Do you have a car that could park itself? Do you let it? Do you trust it? You certainly don't expect it to, yet. However, as time passes, we learn to trust those technologies and expect such functionality. I have to admit that I would be very nervous sitting in an autonomous Uber car. Let alone in an autonomous airplane. However, auto-pilot have been used for quite a few decades. The diffusion of those technologies relies on our willingness to accept them.
Finally, no technology breakthrough was achieved without failure. The more you are willing to try new things and fail, and the more your boss is willing to let you fail--the more successful will those technologies be in changing our lives by 2020.
More on how to forecast technology trends can be found in the 2015 second edition of my book, Bowling with a Crystal Ball: How to predict technology trends, create disruptive implementations and navigate them through industry.