Welcome to the Convergent Science Network Podcast!
During the BCBT Summerschools (2010, 2011, 2012, 2013, 2014) hosting professors Paul Verschure and Tony Prescott interviewed several speakers after their lectures. Interviews are also conducted on other occasions with various scientist in converging fields.
About the hosts: Paul Verschure is ICREA Professor at SPECS, Universitat Pompeu Fabra, Barcelona; Tony Prescott is Professor of Cognitive Neuroscience, University of Sheffield.
The Convergent Science Network of biomimetic and biohybrid systems (CSN, www.csnetwork.eu) is a coordination action for the development of future real-world technologies. CSN is supported through the Future and Emerging Technology programs (FET) of the Information and Communication Technologies (ICT) work programme of Framework Programme 7 of the European Commission.
Audio (post-)production: Sytse Wierenga. Podcast site: Alberto Betella.
Thoughts, discussions, and achievements in neurobiology, biomimetic and biohybrid systems
We can learn a lot from brains and bodies when making machines and robots. But reversely, building complex machine systems can also give ideas about how brains and bodies have implemented their functioning over the evolution of ages. This podcast discusses various themes and aspects in-between robotics, neuroscience, cognitive science, artificial intelligence, biology, and technology.
Interview with Ricardo Sanz
03-04-2018
This post-lecture interview was conducted during the BCBT Summerschool held at the Universitat Pompeu Fabra, Barcelona, september 2010.
The brian is often approached as a control system; as an implemented, crucial strategy to cope with a changing environment through adaptation of behavior. Reversely, for the field of control systems, engineering the brain is an important inspiration of how systems can be developed to deal with dynamic complexity. Ricardo Sanz (Polytechnic University of Madrid, Spain) argues that the required level of complexity for current systems in society has become too high to be controllable by humans: system behavior has become too complex to be analytically solvable, and failures cannot be understood anymore by humans. With Paul Verschure he discusses a strategy to develop and build complex systems that control themselves, and can learn to do so, on inspired and based by some capabilities of the brain like self-awareness, an idea that goes back to classic Cybernetics. In his view, the brain is but one particular example implementation of a system that can cope with changing environmental demands, and that can learn itself how problems in the world could be solved. Control systems are classically not engineered to control themselves, or to be able to adjust to wrong estimates of the controller, instead of wrong estimates of the world (or 'plant'): there is always some human involved for control. But the human brain is not a perfect example, according to Sanz, it is a 'good enough' solution that evolved given the environments of humans; so to copy, or merely mimic the brain is in his view at best insufficient. He theorizes a much higher level of control systems, only partially bio-inspired, that might eventually incorporate elements of self-awareness, but moreover is capable to control processes too complex for current machines, and humans.
About the lecturer
Ricardo Sanz is professor in systems engineering and automatic control and researcher in the field of autonomous systems at the UPM Autonomous Systems Laboratory.
Categories | Autonomous systems | | Robotics
Filetype: MP3 - Size: 36.73MB - Duration: 31:58 m (160 kbps 44100 Hz)
Tweet