0.25
0.5
0.75
1.25
1.5
1.75
2
Learning Dynamic Locomotion Skills for Terrains with Obstacles
Published on Jul 28, 20151544 Views
Using reinforcement learning to develop motor skills for articulated figures is challenging because of state spaces and action spaces that are high dimensional and continuous. In this work, we learn
Related categories
Chapter list
Learning Dynamic Locomotion Skills for Terrains with Obstacles00:00
Movement skills00:27
Physics-based simulation01:51
Results 03:04
High-level Control POlicy03:45
Experience Tuples06:34
Conceptual View 07:28
Value Function Approximation07:29
Policy Approximation10:28
Distance Metric Learning10:43
Outlier Removal10:51
Value Iteration using Positive Temporal Difference Updates - 111:22
Value Iteration using Positive Temporal Difference Updates - 212:11
Local Exploration12:20
Learning Pipeline Overview13:10
Learning Pipeline13:27
Rewards13:45
States14:06
Actions - 114:25
Actions - 214:29
Actions - 314:42
Initial Random-Action Policy14:44
Tight Gaps14:47
Mixed16:07
Continuous Action Space -116:23
Continuous Action Space - 216:36
Comparison of Action Selection Metods17:09
Towards 3D: Quadrupeds17:44
Towards 3D: bipeds17:54
Key Features18:07
Future Work19:21
Questions?19:28