Robotik 2 - Humanoide Robotik, SS19 Karlsruher Institut für Technologie (KIT)

10 |
0:00:00 Start
0:00:05 Imitation Learning
0:11:06 Motivation: Learning from demonstration and human observation
0:19:35 Motion Reproduction using MMM
0:23:06 Open Pose - CMU
0:26:53 Two levels of representation
0:33:09 Segmentation
0:36:10 Key Points
0:41:28 Zero Velocity Crossings (ZVCs)

09 |
0:00:00 Start
0:00:05 Adaptive Haptic Perception = Haptic Exploration
0:02:35 Tactile exploration of a sponge using ARMAR-IIIb
0:04:37 Motivation
0:07:14 What is Haptics?
0:09:47 Haptic Perception
0:11:25 Haptic exploration can only be active
0:13:39 Important questions in haptic perception
0:16:26 Haptic exploration
0:18:21 Potential Field Based Exploration
0:21:15 Exploration using dynamic potential fields
0:25:44 Haptic Exploration with Movemaster
0:27:37 Extracting Grasp Hypotheses
0:30:22 Geometric Filtering and Grasp Computation
0:32:27 Tactile Object Exploration
0:33:34 Combining Vision and Haptics
0:33:54 Examples: Bottle
0:34:30 Visually guided exploration on ARMAR
0:40:06 Combining vision and haptics for grasping
0:45:02 Haptic Exploration for Grasping
0:47:30 Next-Best-Touch for Tactile Exploration
0:50:19 Plan Exploration Actions Efficiently
0:55:12 Next-Best-Touch for Tactile Exploration
0:57:52 Next-Best-Touch
0:58:46 Haptic Exploration for Grasping
1:00:30 How many touches are necessary for grasping?
1:05:24 Exploration and Reconstruction of Unknown Objects using Normal and Contact Sensors
1:08:16 Fundamental question in active hearing
1:09:55 Further research topics

08 |
0:00:00 Start
0:00:08 heutiges Thema: Greifen von unbekannten Objekten
0:01:10 Detecting and Shape Completion Using Symmetry
0:04:31 Shape completion
0:06:19 Finding symmetry planes
0:09:16 Object shape completion
0:11:48 From Low-Level Features to Grasp Hypotheses
0:17:42 ECV based EGA
0:34:03 Training Data for Grasping
0:43:53 Deep learning approaches for grasping
0:48:59 Grasp hypothesis generation and scoring
0:52:56 Generative approach
0:56:58 Deep Grasping
1:08:40 Post Processing
1:11:10 Evaluation
1:15:56 Discriminative approach
1:20:52 Heat map based approach

07 |
0:00:00 Start
0:00:27 Grasping Familiar Objects: Concepts
0:04:21 Grasping Familiar/ Similar Objects: approaches
0:07:41 Discriminative Approaches: Flow-Chart
0:14:57 Discriminative Approaches: Rao et al.
0:19:18 Grasp Synthesis by Comparison
0:29:15 Sensor-Based Exemplars: Herzog et al.
0:39:07 Generative Models for Grasp synthesis: Song et al.
0:45:24 Category-Based Grasping: Madry et al.
0:51:24 Task-Based grasp adaptation
1:12:35 Grasping Objects: Outline
1:17:44 Grasping Unknown Objects: Flow-Chart
1:28:25 Detecting and shape Completion Using Symmetry: Bohg et al.

06 |
0:00:00 Start
0:00:05 Implementation of synergies in Robotics
0:05:58 Mechanical implementation of hand synergies
0:12:10 How to actuate a vector multiple?
0:15:45 How to mecanically add two vector quantities?
0:17:43 Putting everything together
0:20:05 The TUAT/Karlsruhe Humanoid Underactuated Hand
0:33:31 The KIT prosthetic hand
0:39:49 Embedded System
0:54:23 Grasping Objects: Outline
0:56:26 Grasping Known Objects: Typical Flow-Chart
1:03:01 KIT Object database
1:04:30 Grasp Simulator - Simox
1:15:51 Grasping Familiar Objects: Concept
1:21:19 Grasping Familiar / Similar Objects: Approaches
1:24:39 Discriminative Approaches: Flow-Chart

05 |
0:00:00 Start
0:00:17 Bullock Taxonomy
0:04:28 Duality of grasping and balancing
0:11:31 Type of contacts
0:14:11 Taxonomy of whole-body poses
0:15:23 Classification of whole-body actions
0:17:51 Validation of the Taxonomy
0:19:12 Analysis of pose transitions
0:29:23 Language model to generate multi-contact motions
0:33:02 Motion as sequence of poses
0:36:24 Postural Synergies and Eigengrasps
0:36:32 Postural Synergies
0:59:50 Problems with the synergistic model so far
1:01:16 The Soft Synergy Model (1)
1:03:08 The Soft Synergy Model (2)
1:09:30 Eigengrasps