0.25
0.5
0.75
1.25
1.5
1.75
2
GeRT - Generalising Robot Manipulation Tasks
Published on Mar 14, 20123277 Views
The overall aim of the GeRT project is to enable a robot to autonomously generalise its manipulation skills from a set of known objects to previously un-manipulated objects in order to achieve an ever
Related categories
Chapter list
GeRT - Generalising Robot Manipulation Tasks00:00
Consortium00:21
Motivation00:52
Can we reuse these programs?01:46
State of the art01:56
Motivation - 102:48
Key aspects04:05
Integration / Example Programs05:01
What are good example programs?05:25
The tea preparation06:10
Good example programs?06:57
The structure of example programs - 107:43
The structure of example programs - 208:23
Example: Stacking cups08:33
Example: Stacking cups / sensor log09:03
Justin‘s execution environment - 109:09
Justin‘s execution environment - 209:29
Action generalisation / hybrid planning10:48
Action learning overview11:09
Action learning12:00
Geometric Predicate Learning12:54
Using the learned model13:11
Putting it all together13:45
Verfying the plan through hybrid planner15:04
Perception15:17
Perception in GeRT15:33
Shape Similarity: Geometric Feature16:18
Shape Similarity: 4D Feature Density16:26
Shape Retrieval: Example - 116:43
Shape Retrieval: Example - 217:14
Shape Warping17:23
Shape Warping: Alignment17:46
Shape Warping: Correspondence17:53
Shape Warping: Point-Wise18:06
Shape Warping: Grasp Contacts18:28
Learning grasping strategies18:49
Grasping with Motor Primitives18:56
Grasp Adaptation19:07
Imitation by Dynamic Motor Primitives19:11
Results of last weeks integration week19:27
Learning forward models for grasping20:20
Prediction learning for single contacts20:22
Prediction learning & planning with prediction models20:23