BCS-SGAI 2010 Conference Workshop
Bio-inspired and Bio-Plausible Cognitive Robotics
Organisers: Angelo Cangelosi (Plymouth University) and Jeremy Wyatt (Birmingham University)
December 14, Peterhouse, Cambridge
Stream 1, Upper Hall (Peterhouse), 10.30-12.30 and 13.15-14.45
Participants please note that this workshop starts 30 mins before the other workshops
BCS-SGAI Conference Website Directions Registration Printable Programme (pdf)
Bio-inspired cognitive robotics aims at the design of artificial cognitive systems (e.g. humanoid robots, mobile robots) by taking direct inspiration from disciplines investigating behaviour and cognition in natural cognitive systems (humans, other animals). For example, empirical evidence and investigations in cognitive psychology, developmental psychology, behavioural neuroscience and computational neuroscience can directly inform the design of the control systems of robots for tasks regarding vision, action, language and social cognition. In particular, this bio-inspired approach to cognitive robotics is consistent with recent advances embodied in psychology and neuroscience. Both approaches share the idea that in both natural and artificial cognitive systems, cognitive capabilities are intertwined with the s own embodied interaction with the environment. One example is the way that action influences perception in the brain. Bio-plausible cognitive robotics is one way of describing a complementary field that seeks to use robots as to test computational theories of how behaviour is generated in biological systems.
The UK plays a significant leadership role in these fields. This workshop has the aim of providing of overview of UK research in this area and to identify future research challenges and directions in the field.
Topics the meeting will address include: biologically inspired robot control, mechanics and sensing; computational models of motor control, perception, and learning in humans and other animals; robotic implementations of biologically inspired and biologically plausible models; constraints and theories of architectures for biological control, perception, memory and learning.
Key questions include asking what robotics is able to offer studies of cognition in humans and other animals; and what hard knowledge the various tools for understanding human and animal behaviour can offer robotics.
- 10.30 Angelo Cangelosi and Jeremy Wyatt, Welcome
- 10.30 Murray Shanahan (Imperial College), Why aren't our robots as clever as crows?
- 11.00 Barbara Webb (Edinburgh University), What do insects know?
- 11.30 Daniel Wolpert (Cambridge University), Structures and statistics in sensorimotor control
- 12.00 Owen Holland (Sussex University), An anthropomimetic robot: construction, control and cognition
- 12.30 Lunch break
- 13.15 Aaron Sloman (Birmingham University), Genomes for self-constructing, self-modifying information processing architectures
- 13.45 Glyn Humphreys (Birmingham University), Psychological mechanisms of affordance
- 14.15 Yiannis Demiris (Imperial College), Embodied Social Cognition
- 14.45 Final discussion and coffee
Owen Holland, (Sussex University), An anthropomimetic robot: construction, control and cognitionThe idea of investigating human cognition using human-like robots offers three distinct groups of challenges. The first group concerns the design and construction of the human-like robot: which features of the human body should be included, and which can safely be omitted? The second group relates to the necessary control architecture of the robot: how should the conflicts between engineering convenience and bio-inspired implementation be resolved? The final group centres on the core topic of embodied cognition: how could we show that the human-like embodiment contributes to human-like cognition? This talk will discuss these issues in the context of the current EU FP7 project ECCEROBOT (www.eccerobot.org), which takes the view that a truly human-like robotic embodiment requires a much more accurate implementation of the mechanics and dynamics of the human body than has previously been attempted. Such a robot can be thought of as being anthropomimetic, rather than merely humanoid; as will be seen, the potential rewards of this approach come at a high cost in increased practical and theoretical difficulty.
Last updated: Mon 16 November, 2010 Jeremy Wyatt