en
0.25
0.5
0.75
1.25
1.5
1.75
2
Trade-Offs in Robot Skill Learning
Published on Jan 16, 20133044 Views
Related categories
Chapter list
Trade-Offs in Robot Skill Learning00:00
Motivation - 100:04
Motivation - 200:29
Trade-Offs in Robot Skill Learning02:42
Outline - 103:55
Modeling Assumptions05:10
Let the loop roll out!06:46
What is robot learning?06:59
Outline - 207:55
Trade-Off 1: Physical Knowledge vs Generality - 108:05
Trade-Off 1: Physical Knowledge vs Generality - 208:16
Trade-Off 1: Physical Knowledge vs Generality - 308:45
Trade-Off 1: Physical Knowledge vs Generality - 409:28
A Physics-inspired Blue Print - 111:15
A Physics-inspired Blue Print - 212:02
Imitation Learning12:37
Acquisition by Imitation13:16
Trade-Offs in Robot Skill Learning - 114:27
Outline - 314:58
Trade-Off II: Information Loss vs Self- Improvement15:05
Reinforcement Learning15:46
Success Matching16:31
What about the new cost function?18:09
Reinforcement Learning by Reward-Weighted Imitation19:11
Self-Improvement by Reinforcement Learning19:52
More focussed trade-off?21:45
Trade-Offs in Robot Skill Learning - 222:39
Outline - 423:56
Trade-Off III: Large Repertoire vs Parsimonious Behavior Representation24:10
Versatility with Many Primitives24:29
A Physics-inspired Blue Print24:43
Hierarchies with Primitives24:48
Versatility with Many Primitives - 125:02
Need for Separation25:11
Versatility with Many Primitives- 225:38
Localized behavior can be learned efficiently!26:02
Example26:20
Versatility - Parsimony26:34
Demonstrations27:03
Select & Generalize27:16
Self-Improvement28:14
Current Gameplay28:43
Trade-Offs in Robot Skill Learning - 329:14
Outline - 529:31
Conclusion29:34
Thanks29:56