| Abstract |
The purpose of this presentation is to discuss the idea of an "EvoSphere" - new research facility in the intersection of embodied AI, machine learning, robotics, and evolutionary biology. EvoSphere is an integrated platform for multidisciplinary research based on an autonomous ecosystem of self-reproducing and learning artefacts - robots. The (re)production of robots is realized through 3D-printers and semi-automated assembly, where morphology and control both undergo evolution and learning. Not unlike CERN, EvoSphere is complex, expensive, and can be used to carry out research grounded in various disciplines on a time-share basis or, better yet, in an integrated fashion. Further to the ecosystem the facility needs a highly sophisticated apparatus for monitoring, logging, and analyzing huge streams of experimental data, including video and intercepted robot-to-robot communication. Issues in distributed computing and security will be also prominent. EvoSphere also has a great publicity value since live video streams of the ecosystem can be pushed to the internet and we can actively seek a public discussion on the related ethical and philosophical issues. My aim is to acquire an NWO grant for building such a facility supported by a "coalition of the willing" formed by VU and UvA research groups. |