The Service Robot Ecosystem Evolves

Transcript

Trends has been tracking the evolution of "service robots" for many years. As we define them, service robots include self-driving cars, self-piloting airplanes, and autonomous wheelchairs, as well as C3PO-like anthropomorphic robots and simple cleaning robots such as the iRobot Roomba.

Robots have been commonplace in factories for decades, but now, a new generation of robots is also beginning to show up in homes, offices, and hospitals, as well as the battlefield. The more sophisticated robots are looking for snipers, minding children, caring for the elderly, acting as tour guides, and giving people baths or doses of medicine. Less advanced units are performing basic tasks like cleaning floors and mowing lawns. As with all digital technologies, the prices of robots are falling, while at the same time, their capabilities are growing rapidly.

While Moore's law plays an indispensable role, it's not the only really important factor driving the advance of service robots. In fact, it's the "co-evolution" between users' habits and the technological design that is driving the most beneficial results. This "co-evolution" is based on the naturalistic idea that when one organism changes, it influences changes in the other species that interact with it.

In the case of technology, the user changes his or her habits to accommodate the limitations of the technology, and the technology changes to fit the constraints imposed by the user. So ultimately, the user's habits and the technology end up changing many times, converging on an outcome that was not originally anticipated, but that serves a distinct and desired purpose.

An example would be an elderly person changing his or her speaking habits to talk more slowly and precisely to a service robot, because speech recognition software is still evolving. By moving forward without the most sophisticated speech recognition, other, more advanced capabilities of the robot are allowed to evolve and "learn," rather than waiting for the speech component to improve.

Currently, the world leaders in robotic research are Japan and the United States, though some important development work is also coming out of European institutions. There are, however, dramatic differences in the focus of what American and Japanese robots are being targeted to do.

In the U.S., the primary emphasis is on military applications, where robots will be used in place of humans for dangerous battlefield situations. Human soldiers will not be replaced entirely, but the parts of their jobs that have great potential for harm will be carried out by expendable robots. As long as we are engaging in a fight against terrorism, this will be a well-placed priority.

On the other hand, Japan is looking to place robots in a very different role - one that will ultimately become quite common worldwide: that of "caregivers to the aged."

Japan is ahead of the curve in this regard because the country's aging population will increasingly need long-term nursing care. Consider these numbers:

• Nearly 20 percent of the Japanese population is already over 65 — the highest percentage in the world. In rural Japan, it's 25 percent.

• It's projected that the number of Japanese who will need nursing care by 2015 will number 5.7 million.

• Meanwhile, there was a record low of only 16.9 million Japanese children under the age of 15 in 2010.

In a recent issue of The Futurist, Glas observes, "At work, there is no deviation from the established best practice. When I go to the supermarket, they always say exactly the same thing and deliver customer service exactly the same way. So I think the idea of robots doing that sort of work is very natural." This mindset helps explain why Japan has consistently led the world in the adoption of industrial robots.

Adding to this willingness to accept robotic caregivers and other service robots is the Japanese population's aversion to immigrants. This leaves the country without a source of cheap human labor to handle care-giving chores.

When these factors are added up - an aging and decreasing population, an acceptance of automation, and an aversion to immigrant labor - it's clear why Japan will be the leader in humanoid robotic research and development.

Because its needs are so desperate, the Japanese are more likely to accept robot caregivers that are merely "good enough" - that is, at a lower level of sophistication than would be accepted by people in most other affluent nations where there is easy access to immigrant labor.

This dynamic is significant because, as Harvard's Clayton Christensen explains, most disruptive technologies enter the market at the low end, serving those for whom the technology is barely "good enough." In the case of service robots, the lack of acceptable human alternatives in Japan has hospitals, nursing homes, and private residences waiting eagerly for robotic caregivers - even the models that are still in the early stages of development.

In all likelihood, robotic caregivers will follow the usual path of disruptive technologies. Over time, the technology becomes so good that it takes over everything except the highest end of the market. Those companies that aren't in the market early run the risk of being left behind, since the early entrants with the "good enough" solutions will get the opportunity to learn and build brand equity.

Fortunately, American companies have been developing many key technologies as they strive to address military applications and provide simple, low-end household solutions, so much of this expertise can be applied to the care-giving robot market.

1. Ashutosh Saxena, an assistant professor of computer science at Cornell, along with graduate student Yun Jiang, developed a fast and efficient algorithm that uses images from a camera to identify the best places on objects for a robotic arm to grasp. This may sound mundane, yet to do practical, helpful tasks, it will be necessary for service robots to master the ability to pick up objects they have never seen before.
2. Possibly as important will be a robot's ability to "feel" through its "skin" as humans do. Scientists at the Institute of Cognitive Systems at the Technical University of Munich are now developing an artificial skin that will enable a robot to gain tactile information that will enhance its perception of its surroundings and nearby objects when that information is combined with input from other sources, such as cameras and infrared scanners.
3. Researchers at the Georgia Institute of Technology recently focused on how people react to being touched by a robotic nurse - an important issue for the future of healthcare service robots. In their study, people had a generally positive response toward being touched, as long as the intent was to perform a task, such as cleaning, but not to comfort. Interestingly, these results mirror those from similar studies done with human caregivers.
4. Not only will robots need to touch us, their usefulness and effectiveness will be enhanced if they can get "in touch" with us by reading our emotions. Researchers at the Polytechnical University in Madrid, Spain have developed an automated voice analysis program that analyzes the sound measurements of a conversation to deduce human emotion. Another approach is by European researchers who are using artificial neural networks that are adept at reading varied and changing inputs to learn if a person is sad, happy, or angry. Emotion-reading robots are desirable because they will be more readily accepted by the people they serve.
5. To enable placement of robots in unsupervised environments where they will operate independently, engineers at NUI Galway and the University of Ulster are mimicking the neuron structure and operation of the human brain with bio-inspired, integrated circuit technology. Training is similar to the human brain where links between neurons are made and strengthened. Dr. Fearghal Morgan, one of the directors at NUI Galway, explains the researchers' goal: "Our aim is to develop a robust, intelligent hardware neural network robotics controller which can autonomously maintain robot behavior, even when its environment changes or a fault occurs within the robotics system."

Based on this trend, we offer the following forecasts for your consideration:

First, care-giver robots will become "good enough" for the American and EU markets between 2020 and 2025, and that will lead to explosive growth for the industry.

To appreciate how this will happen, consider the tablet computer. Back in the early 1990s, there were rudimentary monochrome tablets with touch-screens running MS-DOS. These used a 2400-baud modem coupled to an analog cell phone to run very job-specific applications.

By today's standards, they were crude and unappealing. Hardly anybody wanted one. But significantly, they offered a huge competitive advantage for those companies that could use them effectively. They were neither high enough quality, nor robust enough in features, for the mass market; they were simply "good enough" for their limited applications.

Decades of consumer experience with laptops and the imagination of Apple eventually paved the way for the mass adoption of the iPad. Similarly, care-giving service robots are likely to be "good enough" for many Japanese institutions and homes when they become reliable enough for widespread deployment; perhaps as soon as 2015.

Once they are deployed in the hundreds of thousands due to Japan's adoption, prices will plunge and capabilities will soar. Nevertheless, it will likely be 2025 before they become "good enough" to replace human caregivers in the West. Yet, when that tipping point arrives, expect the breakthrough entrant to become the care-giving robotic equivalent of the iPad.

Second, widespread use of service robots will depend on the development of the supporting environment needed to fully utilize them.

In the next 15 years, service robots will integrate into a broader web of technology that we at Trends call the "Internet of Everything." Within this tiered network, all devices will be intelligently linked to one another through a backbone of wireless nodes that will receive, analyze, and transmit streams of data.

The so-called "smart home" is at the heart of this idea. Within this paradigm, systems throughout the home, such as heating, air conditioning, refrigerators, multimedia centers, and lighting will be automatically regulated to meet specific demands. This will extend all the way down to RFID tags on nearly every item in the closet or refrigerator.

The network will even integrate the residents' clothing, which will monitor their health. Service robots will function within that environment, interacting with the people, as well as the house and its contents. Outside the house, GPS and other navigational aids will enable robots to function effectively. The service robots and their ecosystem are likely to show up first in hospitals, nursing homes, and assisted-care facilities; but once proven, they will transition rapidly into private homes.

To make this vision cost-effective requires little more than the steady advance of Moore's law and a parallel increase in wireless system performance over the next 15 to 20 years. While breakthroughs are clearly necessary in the software world, much of the fundamental research is already underway in Japan, the U.S., and the EU.

Third, it's possible that the growing "skills mismatch" will delay widespread adoption of service robots.

Previously in Trends, we've discussed the growing skills mismatch in the human labor market: a severe shortage of high-skilled technologists, coupled with a surplus of workers with a high school diploma or less. (This situation is further exacerbated by the surplus of college graduates with degrees in fields that have a low technology content.)

To develop service robots and an industry to support them, we'll increasingly need people to design, build, program, and monitor these robots. Meanwhile, widespread adoption of robots will eliminate many jobs that are now filled by relatively unskilled workers, while creating higher paying jobs for highly skilled workers. Because of the skills gap, these new jobs will be slow to fill, potentially delaying the emergence of service robots.

Comments