Shaping in Practice: Training Wheels to Learn Fast Hopping Directly in Hardware
| Autor: | Alborz Aghamaleki Sarvestani, Alexander Spröwitz, Felix Ruppert, Steve Heim |
|---|---|
| Rok vydání: | 2017 |
| Předmět: |
FOS: Computer and information sciences
0209 industrial biotechnology business.industry Computer science Robotics 02 engineering and technology Robot leg Computer Science - Robotics 03 medical and health sciences 020901 industrial engineering & automation 0302 clinical medicine Robustness (computer science) Robot Artificial intelligence business Robotics (cs.RO) 030217 neurology & neurosurgery Computer hardware |
| Zdroj: | ICRA |
| DOI: | 10.48550/arxiv.1709.10273 |
| Popis: | Learning instead of designing robot controllers can greatly reduce engineering effort required, while also emphasizing robustness. Despite considerable progress in simulation, applying learning directly in hardware is still challenging, in part due to the necessity to explore potentially unstable parameters. We explore the concept of shaping the reward landscape with training wheels: temporary modifications of the physical hardware that facilitate learning. We demonstrate the concept with a robot leg mounted on a boom learning to hop fast. This proof of concept embodies typical challenges such as instability and contact, while being simple enough to empirically map out and visualize the reward landscape. Based on our results we propose three criteria for designing effective training wheels for learning in robotics. A video synopsis can be found at https://youtu.be/6iH5E3LrYh8. Comment: Accepted to the IEEE International Conference on Robotics and Automation (ICRA) 2018, 6 pages, 6 figures |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|