Morphological Computing in Machines and Animals
Abstract
To develop a new synthetic approach to morphological computation, we propose to advance our conceptual framework of Templates and Anchors. To begin immediate testing of our framework of defining control affordances of complex morphologies, we will explore our already available data on animal and legged robot maneuvers. We will begin collecting data to examine the problem of terrestrial self-righting in animals and machines as a specific domain of legged locomotion tasks that is simultaneously of great interest to biologists and would confer substantial benefit for robotics could some of the impressive animal capabilities in this realm be transferred to our machines. At the same time, because it is clear that the animals of interest use their bodies with great mechanical as well as neural intelligence in accomplishing these complicated maneuvers, the task domain is particularly rich, but tractable, ground for advancing insight into morphological computation. With aspirations to understand the most complex animal behaviors and implement them in machines, we will begin to examine data we collected on more agile animals, as they point the way to our grand challenge of using morphological reduction and compositions of simple dynamic models to advance morphological computation.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Oct 16, 2018
- Source ID
- W911NF1710229
Entities
People
- Daniel E. Koditschek
Organizations
- Army Contracting Command
- United States Army
- University of Pennsylvania