Shaping Behavior With Adaptive Morphology
Martin Garrad, Jonathan Rossiter, Helmut Häuser
- Year
- 2018
- Citations
- 17
- Access
- Open access
Abstract
Robots that exploit their embodiment promise to be more robust, energy-efficient, and adaptable. However, the majority of systems designed in this way are only capable of exploiting their embodiment when performing a single task in a single environment. For such robots to be capable of adapting to a range of tasks or environments, they must be capable of adjusting their morphology online and have an understanding of how adjustments in both control and morphology affect their behavior. We introduce the concept of the control-morphology (CM) space and the variable stiffness swimmer, a multisegment pendular robot with adaptive joint stiffness. This system allows us to perform an initial investigation into how navigating the CM space affects the behavior of a robot. We show that the behavior of the system is not only the result of selecting a particular location within the CM space, but also the route taken to arrive at that point. We also demonstrate locations within the space where a shift in behavior can be caused entirely by smooth online changes in morphology.
Keywords
Related papers
Statistical Learning Theory
Yuhai Wu, Vladimir Vapnik
1999
Artificial intelligence: a modern approach
1995
Fractional Differential Equations
Igor Podlubný
2025
Applied Nonlinear Control
Jean-Jacques Slotine, Weiping Li
1991