Distributed Learning for Complex Policies
Abstract
The modular snake robots in Carnegie Mellons Biorobotics lab provide an intriguing platform for research: they have already been shown to excel at a variety of locomotive tasks and have incredible potential for navigating complex terrains, but much of that potential remains untapped. The motivation to expand the capabilities of these robots stems from experiencing several failures and limitations in real world tests. For example, this robot was able to navigate a narrow passage underneath a rubble pile at a disaster response training site, but was unable to pass over a four inch high piece of wood which lay across its path once the passage widened. In an archaeological expedition near the Red Sea, the robot was able to move further than a human could into a collapsed cave containing four-millenia-old ship timbers. However, a gradual sandy slope prevented the robot from moving further and potentially making an archaeological discovery. To extend the capabilities and learn improved controllers for these robots, a model must be found on which to evaluate the controller. Unfortunately, analytic snake robot models have proven overly complex due to the robots complex, multi-modal locomotion dynamics, and after pages of formulae these models only provide qualitatively accurate results.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jun 29, 2019
- Accession Number
- AD1091006
Entities
People
- Howard M. Choset
- Jeffrey Snyder
Organizations
- Carnegie Mellon University