A Tipping Point in Robotics, and My Five Predictions for 2016
While scientists are making progress in understanding how the mind works, robotics has failed to live up to the expectations set in the 1950s and ‘60s for human-like robots, exemplified by the Jetson’s cartoon character, Rosie the robot maid.
Since the rejuvenation of robotics began in the 1980s, we have discovered that engineering and building a robotic “body” is attainable. But it is the computer-based artificial brain that is still well below the level of sophistication needed to build an intelligent, human-like robot.
In this pursuit of intelligent robots, waves of researchers have grown disheartened and scores of start-up companies have gone out of business.
Researchers in the field have found that it has been easier to mechanize many of the high-level cognitive tasks we usually associate with human “intelligence,” such as symbolic integration, proving theorems, playing chess (refer to IBM Deep Blue), competing on the Jeopardy! TV game show (refer to IBM Watson) television show, or diagnosing a medical condition by using a Bayesian Network.
However, it has been very hard to mechanize the kind of tasks that many animals can do easily, such as:
—Walking around without running into things (obstacle avoidance).
—Catching prey and avoiding predators (behavior-based control).
—Interpreting complex sensory information (perception).
—Modeling the internal states of other animals from their behavior (state estimation).
—Working as a team (swarms).
So, is there a fundamental difference between the two categories of (a) high-level cognitive tasks and (b) animal tasks?
High-level cognitive tasks include reasoning, learning, and inference. Animal tasks are more behavior-based, and complement high-level thinking. Both categories are progressing steadily, taking advantage of the increased computer processing available. History has shown that implementing the simple behaviors of animals have proven difficult. In a different way, implementing the higher thought logic of the human brain has also proven difficult without having a proven existing model. The maturation of the two categories, and their integration, will lead towards a general unified theory of intelligence in the coming years.
So what can we expect to see in 2016? Here are my five predictions in robotics for the coming year:
—The explosive sales of drones will continue, as regulatory rules lag their implementation—unless there is a drone-related airliner disaster or near-disaster. FAA regulations have been a long time coming, and are expected to be released in the fourth quarter of 2016. Currently, commercial operations of drones under 55 pounds are prohibited unless operators get a Section 333 exemption, a Certificate of Authorization, and have an air-worthiness certificate. The new regulations will clear up the rules.
—Commercial investments will continue but many foundational investors will be cautious as they survey the market for robotics. Companies like Qualcomm (NASDAQ: QCOM) become case studies for the emerging robotics industry as their principle businesses hit choppy waters, with the ripple effect restricting its investments in the field. Robotics have been predominantly a research field that was funded by the military, but in the past few years, there has been an abundance of robotic start-up companies. These new companies are entrepreneurial in nature and still unproven.
—Although major robotics efforts are already underway by Amazon, Google, Facebook, Toyota, Sony, Bosch, DHL (parcelcopter), Uber, and others, new international players will enter the market.
—New advances will be made in Artificial Intelligence by such forward-thinking companies as Google’s DeepMind Technologies and Baidu’s internal research activities.
—Industrial automation will continue to expand in East Asian factories, and new adjacent markets will emerge in bio-robotics, healthcare (including pharmaceuticals), retail kiosks, and legal analysis and processing.
Whatever happens in 2016, a tipping point of unstoppable change already is occurring in robotics, brought about by the impact of Moore’s Law on communication, computation, sensing, and control.
Robots are complex systems, benefiting from applications using probabilistic calculus to handle uncertainties and new Artificial Intelligence techniques in learning and reasoning. We are getting ever closer to achieving some of the robot promises that were dreamt about in the 1950s.
Current robot applications are quickly expanding beyond the industrial and manufacturing sectors that have supported the technology in the past. There is an emerging hobby and service industry market in the U.S. with drones and multi-rotor aircraft.
At the same time, there are other efforts in robotics progressing in Asia, Europe, the Middle East, and throughout the U.S. For the government (in both military and non-military applications), autonomy is a key capability that will continue to drive innovation. Hardware costs will continue to go down and the growth in processor consumption will usher in greater interest in multi-agent and distributed robotic systems, leading ultimately to robot swarms.