some very very basic and or old thoughts
play

Some Very, Very Basic and/or Old Thoughts on Multimodality and - PowerPoint PPT Presentation

Some Very, Very Basic and/or Old Thoughts on Multimodality and Uncertainty Oliver Brock Robotics and Biology Laboratory The Amazon Picking Challenge The Problem Possible Reasons for Winning Luck Bigger team Graduate students


  1. Some Very, Very Basic and/or Old Thoughts on Multimodality and Uncertainty Oliver Brock Robotics and Biology Laboratory

  2. The Amazon Picking Challenge

  3. The Problem

  4. Possible Reasons for Winning… ► Luck ► Bigger team ► Graduate students versus undergraduates ► Cheating ► Embodiment ► Vertical integration (behavior) ► Better boxes: Factorization ► Embodiment ► Biases / priors / heuristics ► Project management ► Really smart people

  5. Embodiment Choices

  6. Boxes

  7. VISION

  8. Prior Knowledge for Segmentation

  9. Projecting to 1D Color (Hue + B/W) Hue + white/gray/black

  10. Probabilistic Segmentation P(Color | Object) Backprojection Bayes' rule

  11. More Features! Color Height (3D) Edges Height (2D) Combine Segment 3D missing Distance to shelf

  12. APC Segmentation Results

  13. Importance of Each Feature

  14. James J. Gibson (1904 – 1979) Eleanor J. Gibson (1910 – 2002) 1966

  15. Prog. by Demo Prog. By Demo Kinematics Surveillance Activity Objects Manipulation

  16. “Learning to See”

  17. Experimental Design 576-dimensional observation 10/10/2015

  18. Training (~ 50 minutes)

  19. Test of Learned Behavior

  20. What the Robot Sees and How It Learns

  21. Learning to See robotics prior state reinforcement learning action 10/10/2015 32 Rico Jonschkowski

  22. Learning to Perceive

  23. Learning to Perceive

  24. Moving in An Environment With Disturbances

  25. Moving in a “Cluttered” Environment (Robot’s View)

  26. Performance

  27. Different Learned Representations

  28. How Do We Do It? Rico Jonschkowski William of Ockham Sir Isaac Newton 1287 – 1347 1642 – 1727 1987 –

  29. Five Robotic Priors Simplicity Only a small number of world properties are relevant. Task-relevant properties of the world change gradually. The task-relevant properties together with the action determine the reward. The amount of change in task-relevant properties resulting from an action is proportional to the magnitude of the action. The task-relevant properties and the action together determine the resulting change in these properties.

  30. Functional Relationships between DOF

  31. Online IP: Three Recursive Estimation Problems spatial coherency rigid body physics kinematics feature motion rigid body motion kinematic model measurement input input to process model

  32. Visual Odometry and Online IP

  33. spatial coherency rigid body physics kinematics feature motion rigid body motion kinematic model rigid body motion shape reconstruction (shape) rigid body physics shape appearance

  34. Effect of Integrated Tracking on Reconstruction Integrated Tracking Non-integrated Tracking

  35. Robotic Senses Considered as a Perceptual System PRIOR PRIOR PRIOR ESTIMATION ESTIMATION ESTIMATION PRIOR PRIOR PRIOR ESTIMATION ESTIMATION ESTIMATION PRIOR PRIOR PRIOR ESTIMATION ESTIMATION ESTIMATION

  36. Layers of the Cortex / Connectivity of Visual Cortex

  37. Uncertainty: the Ruler of Perception

  38. Roberto Martín Martín Sebastian Höfer

  39. Our Options Against Uncertainty ► Ignore it! Assume you know everything! (Ignore this.) ► Model it and then go back to the above ► Let the physics of interaction deal with it ► Stay in regions of the space (what space?) where uncertainty is not (that) relevant to task success

  40. Modeling it and then Go Back to the Above Assumption: Only the world is a truthful and complete model of itself.

  41. Explicit Reasoning about Uncertainty cost of modeling benefit/cost and computation High Low no uncertainty as much uncertainty as possible

  42. Let the Physics of Interaction Deal with It

  43. Pisa/IIT Soft Hand

  44. How to Identify Regions of Robust Interaction? This slides was intentionally left blank.

  45. Clemens Eppner Raphael Deimel Jessica Abele

  46. Conclusion How? What? PRIOR PRIOR PRIOR ESTIMATION ESTIMATION ESTIMATION Don’t try to model the world! PRIOR PRIOR PRIOR Instead, try to model robust, ESTIMATION ESTIMATION ESTIMATION task-relevant correlations in action and perception space. PRIOR PRIOR PRIOR ESTIMATION ESTIMATION ESTIMATION INTERACTIVE PERCEPTION

Recommend


More recommend