Self-learning robots: indispensable friend or threat?

alibaba

Self-learning robots: indispensable friend or threat?

16-12-2015 | Insight | Henk Grootveld, Marco van Lent

Although some see robots as a threat, they may also provide a potential solution to current problems, such as the lack of care for an aging population. This was the idea behind a lecture by Pieter Jonker, professor of Vision-based Robotics at Delft Technical University, in honor of the 50th anniversary of the growth fund Rolinco.

Speed read:

• Robots can help out in the healthcare sector
• Jobs disappear, but new ones are created
• Rolinco has varied robotics exposure in its portfolio
  

How do you make an intelligent robot?
It seems so simple: a robot butler that brings you a cup of coffee, or a robot security guard in a factory that detects a missing pallet. But such robots first require a great deal of production, programming and endless learning processes. To function intelligently and autonomously, a robot must first master cognitive learning. In practice this means that it recognizes what someone is doing, such as serving a cup of tea, and what that means. But it must also learn skills through trial and error, such as walking and picking up objects.

Making an intelligent autonomous robot requires a software architecture of no less than 17 layers with the aim of designing and controlling all its functions. One of the basic layers is to program the robot to guarantee its own safety. The robot must then be capable of fine motor skills that it can refine and practice. More difficult still is to teach a robot intuitive behavior, but it really begins to get high-tech when trying to program a robot to develop the desire to learn. One of the last stages is to teach the robot communication skills and emotional interaction and even to develop a capacity for intROSpection and ethics.

Practical applications
It may sound like pie in the sky, but there are already intelligent robots suitable for use in both industrial and domestic settings. Professor Jonker has for instance helped to develop a robot that guards warehouses. This robot detects if a machine is missing, alerts a control center, even drawing up the necessary report itself. It can also detect people and confirm whether they are permitted to be there, and if not, to alert security. It is fitted with sensors for gas or heat detection.

Similar robots already operate in the Port of Rotterdam to detect whether a ship is permitted to dock, or if a container is being broken into. And there are even robots in the pipeline for underwater inspection. The challenge here is that GPS does not work underwater and systems cannot communicate with each other.


In terms of mobility, people movers are being developed as a sort of autonomous taxi, fitted with cameras and sensors. In the spring of 2016, one of these driverless cars is set to drive from Ede-Wageningen station to the campus of Wageningen University.

Care services
One practical application that could be of major benefit to an aging population is the care robot. Such robots can carry out simple procedures, reducing the number of bedside hands needed. A human care provider could operate various robots remotely. Robot EVA can recognize faces and emotions. Paro the hug robot, which can respond with emotions, is already helping to care for dementia patients in Japan.

LEA (the Lean Elderly Assistant walker) is due to be launched in 2017. LEA has been created to help people stay in their own home for longer when they suffer mobility or cognitive problems, for instance, or struggle with muscle tremors. LEA can be fitted with a protocol for each day. Many elderly people suffer from memory loss, forgetting how to switch a kettle on, for example, but LEA is able to instruct them. Other intelligent robots are able to detect if someone falls over.

Disappearing jobs
Hazardous, dirty or boring jobs could be done by robots in future. For example, meter readers, cashiers, taxi drivers, but also tax inspectors and legal assistants. Occupations still requiring human input include not only jobs for the well-educated, such as lawyers or doctors, but also jobs that involve creativity, artistic sensibilities and care, such as hairdresser, plumber, preschool teacher, fashion designer or therapist.


But a bad teacher leads to a poorly functioning robot. And if a robot takes a wrong decision, it must be able to account for its behavior, perhaps even before a judge. These new self-learning robots will therefore generate demand for specialized occupations, such as teachers, communication specialists, and psychologists for robots

With the emergence of autonomous, self-learning robots printed cheaply using 3D technology, simple services will in future cost next to nothing. As a result, self-learning systems are set to play an increasingly important role in our lives.

What are we going to do with all the free time?
As robots carry out more and more tasks for us, even driving cars, this will create a greater amount of free time. So what will we do with that time? Artist Sander Veenhof expects that we will spend some of it on ‘augmented reality’. In 2016, the Oculus Rift will be released, a virtual reality glass that allows you to construct a Minecraft structure in your living room, or take a virtual ride in a rollercoaster.

  

‘Lazy’ travelers will no longer have to visit the Great Wall of China – just pop on your Oculus Rift and walk up some stairs to enjoy the sensation. A cheaper imitation version of this headset is already available, consisting of a cardboard box costing 5 euros in which you mount your smartphone. But Veenhof takes it a step further and makes all sorts of applications to virtually change reality. You can for instance hold your smartphone against a wall to view the most fantastic paintings on that wall. Or you see someone sitting at a table in a restaurant and can view what he or she is tweeting – or could be tweeting. And even dating is no longer sacred: Veenhof has already done some experiments with a sort of ‘cyborg dating’, with one person wearing an augmented reality headset and the other guiding them by the hand and giving instructions.

Rolinco and robotics
Within the ‘Industrial Renaissance’ trend, Rolinco invests in new technologies, such as robotics. The fund has exposure to Fanuc in Japan and to Swedish-Swiss ABB. Both are world leaders in the production of industrial robots and so-called collaborative robots – small robots that can cooperate with people. Additionally, Rolinco invests in Sensata in the US, which makes sensors for robots and other devices