FoCAS sponsor best presentation award at Workshop on Nature-inspired techniques for robotics

Best student presentation

Jorge Gomes (centre) receives the FoCAS best student presentation from NIR workshop organisers Claudio Rossi (Universidad Politécnica de Madrid, Spain)
 and Nicolas Bredeche (University Pierre et Marie Curie, France)

NIR 2014: September 13 2014, Ljubljana, Slovenia

Jorge Miguel Carvalho Gomes is a second year student at University of Lisbon who won the best student presentation award sponsored by FoCAS at the Nature-inspired techniques for Robotics workshop (NIR 2014). This took place on 13 September at the 13th international conference on Parallel Problem Solving from Nature (PPSN) in Ljubljana, Slovenia. Thirteen presentations covered nature-inspired approaches to robotics including collective adaptive systems, evolutionary robotics, bio-inspired design and control.

Jorge Gomes is a second year PhD student at the Universidade de Lisboa in Portugal associated with two labs: LabMAg (Laboratory of Agent Modeling), and Instituto de Telecomunicações. His presentation was entitled “Diversity-based Coevolution of Behaviourally Heterogeneous Multi-robot Systems” developed with P Mariano and A L Christensen, and the abstract is below

Heterogeneity is considered fundamental for effective task execution in several complex, real-world scenarios. Evolutionary computation has the potential to automate the design of multirobot systems, but to date, it has mostly been applied to the design of homogeneous systems. In our research, we investigate how behavioural diversity techniques can help overcoming two of the main issues in cooperative coevolution: premature convergence to stable states, and poor scalability with the number of agents. So far, we have showed that driving coevolution with behavioural novelty measures can mitigate the problem of premature convergence. In ongoing work, we are studying how the scalability problem can be addressed, by allowing the emergence of partially heterogeneous teams, based on the diversity of agent behaviours observed within a team.

The purpose of the NIR 2014 workshop was to analyze the state-of-art / state-of-knowledge in the field of nature based computation techniques to automatically design robot controllers and/or morphology, as well as learning and optimization of real or simulated autonomous robots. This workshop aimed to bring together researchers and experts in different disciplines, fostering interdisciplinary debate between fields such as neuro-evolution, evolutionary design, artificial life, evolutionary robotics, developmental and learning in robotics, adaptive collective robotic systems, etc.

Visit Workshop website