Artificial
Autonomy

How far are we really getting to the goal?
Is artificial autonomy possible?

AlifeX Workshop on Simulation Models of Autonomous Systems

In June 2006 the 10th Artificial Life Conference will take place in Bloomington (Indiana University, USA). In the context of that conference we are organizing a workshop on artificial autonomy. The main idea would be to evaluate, 15 years after the proposal of Bourgine and Varela (1991) for a “Practice of Autonomous Systems”, how far has the field moved towards that goal.

PROPOSAL

Organizers

Xabier Barandiaran and Kepa Ruiz-Mirazo

Extended description of the workshop

In the Proceedings of the First European Conference on Artificial Life [Varela & Bourgine (1991) Towards a Practice of Autonomous Systems. Proc. of the First ECAL, pages xi—xvi] a full research program on the artificial implementation of autonomous systems was envisioned. According to that view, autonomy was the fundamental property underlying both the phenomenon of life and cognition, so the challenge of the new artificial sciences (if they were to illuminate our understanding of those phenomena) consisted in the simulation or realization of systems with increasingly autonomous capacities. In other words, artificial autonomy was taken up as a difficult but achievable research goal, whose pursuing could, furthermore, contribute to merge the new and old paradigms of Artificial Life and Artificial Intelligence and to uncover some of the most elusive properties of biological organization.

More recently, Francisco Varela's passing away triggered a number of publications, conferences and workshops somehow reinforcing the research agenda on issues dealing with autonomy and autopoiesis at different scales (origins of life, immune system and neural dynamics, ...). At the same time recent advances in neurosciences and robotics, as well as in cellular modeling techniques have significantly contributed to the effort.

More recently, Francisco Varela's passing away triggered a number of publications, conferences and workshops, somehow reinforcing the research agenda on issues dealing with autonomy and autopoiesis at different scales (origins of life, immune system and neural dynamics, ...). At the same time recent advances in neurosciences and robotics, as well as in cellular modeling techniques, have significantly contributed to the effort.

Therefore, we consider that it is a suitable time to evaluate the efforts made so far and try to assess to what extent the goal of an artificial autonomy is actually within reach or constitutes just a philosophically utopian quest. In order to do so, we propose a workshop focused on recent (if possible new, original) simulation models of autonomous systems, coming from the fields of biological and cognitive sciences (i.e., based on artificial chemistries as well as on artificial neural networks or dynamic sensorimotor controllers).

Expected contributions

Our main interest is to check the present 'state of affairs' regarding the simulation of autonomous systems, both in a basic biological sense and in the more elaborate cognitive domain. Therefore, original simulation models of this kind of systems will have priority in the selection process. More specifically, we would like to encourage researchers working on artificial chemistry models that tackle the problem of implementing minimal (cellular) metabolisms, or on neural network models that approach cognition from a minimal but embodied perspective. Other types of contribution (like novel results in related disciplines that can inform the debate or review articles on the general subject of artificial autonomy) will also be welcome, although the space given to the latter will depend on amount and quality of the former.

Workshop Format

Oral contributions will be grouped in three different sessions: (i) basic autonomy - origins of life/metabolism, (ii) cognitive autonomy and (iii) historical-philosophical perspectives. Presentations are expected to be around 15-20 min. long, with half-an-hour round-table discussion at the end of each session. The contents of the talks will be distributed in advance (both as previously downloadable files in the workshop website and as hand-copies in situ) so that we engage in more interactive and productive discussions. The final part of the workshop will be dedicated to establish a working definition of 'natural autonomy' (based on a previous draft), which would be aimed as a contribution to wikipedia, and to reach a minimal consensus on the research roadmap for 'artificial autonomy'. After the conference the website will be updated with the main results of the workshop and will remain open for further discussion and contributions.

Registration

In order to take part in the workshop registration for the full conference is required (no additional registration needs to be done).

Program Committee

Alvaro Moreno, Arantza Etxeberria, Barry McMullin, Cliff Hooker, Evan Thompson, Ezequiel di Paolo, John Stewart, Jesús Siqueiros, Jon Umerez, Juan Carlos Letelier, Paul Bourgine, Pier Luigi Luisi, Randall Beer, Takashi Ikegami.