How do we stop ‘robot evolution’ from spiraling out of control? Two essential articles on the future of robotics and AI

By Colm Gorey, Frontiers Science Communications Manager


Robotics and AI have quickly outgrown as tropes of science fiction and are almost ubiquitous in our daily lives, even if we don’t realize it. Now, Frontiers highlights just some of the latest top articles in this fascinating field.

Whether it’s wheeled robots whizzing around city-sized warehouses, or if it’s just a smart assistant in your phone, robotics and AI has quickly become integral technologies in our lives. And while big tech companies and gadget lovers are quick to speak of its enormous benefits, those closely studying the field are calling on more caution. These technologies are working well now, but can we allow them to develop indefinitely?

That’s among the questions being asked by some of the largest governmental bodies in the world – such as the EU and UN – who believe stricter rules and regulations are needed to keep technology not only within the realms of rapidly-outdated laws, but make sure they don’t eventually lead to our downfall.

To that end, here are some of the latest articles published to Frontiers authored by some of the top researchers in their field as part of the research topic ‘Horizons in Robotics and AI’.

Robot Evolution: Ethical Concerns

We might often think of a robot as something which is built and exists until it either breaks or is decommissioned. But can a robot truly evolve, and if so, how can the human race prevent it from ushering in our extinction?

That’s the question posed by a team from the Netherland and the UK in a paper published to Frontiers in Robotics and AI. It pointed out that until now, work on evolutionary robotics has mostly been performed in computer simulations, safely confined to a virtual world inside. Occasionally, the best robots in the final generation have been constructed and materialized in the real world, but even in these cases the evolutionary process itself took place in simulation.

Some studies have demonstrated self-reproducing physical machines, but the resulting system was not evolutionary because there was no inheritance and reproduction created identical clones without variation. However, recent advances robotics, 3D-printing, and automated assembly are enabling systems of physical robots that can autonomously reproduce and evolve.

The authors’ key message is that awareness must be created before the technology becomes mature and researchers and potential users should discuss how robot evolution can be responsibly controlled. Specifically, robot evolution needs careful ethical and methodological guidelines in order to minimize potential harms and maximize the benefits.

Article link:

Robot Learning From Randomized Simulations: A Review

While AI has advanced rapidly to such an extent that various programs can fool many into thinking they’re speaking with a human, robots remain oddly clunky and far from reaching their science-fiction like potential.  

In a review article published to Frontiers in Robotics and AI, an international team of researchers pointed to the fact that he rise of deep learning has caused a paradigm shift in robotics research, favoring methods that require large amounts of data.

Yet it is prohibitively expensive to generate such data sets on a physical platform. Pondering how this ‘reality gap’ can be overcome, the researchers aimed to find out how simulators can be modified to facilitate learning robot control policies and overcome the mismatch between simulation and reality.

Article link:

REPUBLISHING GUIDELINES: Open access and sharing research is part of Frontiers’ mission. Unless otherwise noted, you can republish articles posted in the Frontiers news site — as long as you include a link back to the original research. Selling the articles is not allowed.

%d bloggers like this: