Evolvable Morphologies

There are only so many tasks which, desirably, would be performed by robots instead of humans. Beyond that, there are also tasks that simply could not be realized by humans, as for instance when exploring inhospitable locations. When designing a robot to solve a well-defined problem, it would possible to precisely engineer all of its traits. However, in some cases, like the aforementioned one, the problem itself is not totally known, making it impossible to construct such robot. In this content, a system that could be deployed in the expected environment, and then, autonomously design the appropriate robot for the due task would be a solution to this engineering endeavor. The long-term vision behind this project foresees the creation of populations of robots that could evolve without human intervention. For a system like that to succeed, there are uncountable challenges to be met. And, what we envision, is to improve the current state of the Evolutionary Robotics field of research, by studying the benefits of having robots which would have their bodies and brains, and thus their behavior, regulated during life-time according to the environment they are in at each moment. Having in mind that a robot may have to go thorough different areas or states of one unknown environment, this regulation would allow robots to be more robust. The inspiration for such regulation process comes from nature, and refers to phenotypic plasticity, which can be observed in nature.

Team

Karine Miras
PhD Student

Karine Miras

Prof. dr. Guszti Eiben
Head of the Group

Prof. dr. Guszti Eiben