Probabilistic Online Prediction of Robot Actions Results based on - - PowerPoint PPT Presentation

▶
probabilistic online prediction of robot actions results
SMART_READER_LITE
LIVE PREVIEW

Probabilistic Online Prediction of Robot Actions Results based on - - PowerPoint PPT Presentation

Aug 13, 2023 287 likes •619 views

Probabilistic Online Prediction of Robot Actions Results based on Physics Simulation Functional Imagination Sebastian Rockel TAMS, University of Hamburg, Hamburg Nov. 30, 2015 Journal of Economic PerspectivesVolume 29, Number

slide-1
SLIDE 1

Probabilistic Online Prediction

  • f Robot Actions Results

based on Physics Simulation

Sebastian Rockel
 
 TAMS, University of Hamburg, Hamburg Nov. 30, 2015

”Functional Imagination“

slide-2
SLIDE 2

Journal of Economic Perspectives—Volume 29, Number 3—Summer 2015—Pages 51–60

Is a Cambrian Explosion Coming for Robotics? Gill A. Pratt

Eight Technical Drivers 1. Exponential growth in computing performance 2. Improvements in electromechanical design tools and numerically controlled manufacturing tools 3. Improvements in electrical energy storage 4. Improvements in electronics power efficiency 5. Exponential expansion of the availability and performance of local wireless digital communications 6. Exponential growth in the scale and performance of the Internet 7. Exponential growth of worldwide data storage 8. Exponential growth in global computation power

2

slide-3
SLIDE 3

Journal of Economic Perspectives—Volume 29, Number 3—Summer 2015—Pages 51–60

Is a Cambrian Explosion Coming for Robotics? Gill A. Pratt

3

Cloud Robotics: Big Idea #3: Learning from Imagination

slide-4
SLIDE 4

Outline

  • 1. Related Work
  • 2. Architecture & Integration
  • 3. Uncertainty
  • 4. Experiments & Results
  • 5. Conclusion

4

slide-5
SLIDE 5

Task Planning

5

symbolism continuity

Table

x,y,z discreteness

Table

x,y,z x,y,z x,y,z x,y,z x,y,z

Table

…

slide-6
SLIDE 6

Functional Imagination

  • improve task-planning based system
  • use prediction from physical simulation

6

  • integrate prediction
  • use predicted results to adapt


plan execution

  • forestall failures
slide-7
SLIDE 7

Related Work

  • concurrent reactive planning (Beetz 2000)
  • meta-CSP: hybrid planning/reasoning (Moffit et al.

2006)

  • functional imagination (Marques et al. 2008)
  • GTP + HTN (de Silva et al. 2013)

7

slide-8
SLIDE 8

(2012-2015)

  • ”Robustness by Autonomous

Competence Enhancement“

  • high-level world representation
  • multi-level experience representation
  • learning and generalizing from

experiences

8

slide-9
SLIDE 9

RACE Architecture

9

Conceptualizer Reasoning

new concepts fluents plan ex− data fluents fluents periences sensor actions ROS plan fluents plan, goal action results OWL concepts

Experience Extractor/ Annotator

instructions

Memory

ROBOT Perception User Interface HTN Planner

periences ex− instructions data continuous concepts initial state, goal schedule fluents, OWL

Monitor Execution Semantic

Capabilities Perceptual

Blackboard

OWL Ontology Interpretation and

slide-10
SLIDE 10

Methodology

10

Sense Act Plan Sense Act Plan Imagine

hierarchical

slide-11
SLIDE 11

Methodology (cont’d)

11

Plan Sense Act Plan Sense Act Imagine

reactive

slide-12
SLIDE 12

Processing

12

plan generation execution robot projection

is
 imagine
 action?

simulation evaluation yes no

slide-13
SLIDE 13

Architecture

13

Planner Black- board Executor Robot Capabilities plan fluents ROS actions Functional Imagination Simulation 1 Simulation 1 Simulation 1 fluents imagine
 actions imagine
 actions results

slide-14
SLIDE 14

Architecture (cont’d)

14

Functional Imagination Evaluation Translation Imagination Client 1 (PC 1) Simulation 1 Blackboard Executor Imagination Client 1 (PC 1) Simulation 1 Blackboard Executor Imagination Client 1 (PC 1) Simulation 1 Executor Black- board

slide-15
SLIDE 15

Plan Adaptation

15

act1

execution time imagination

par1 par2 par3

act<..>

… …

slide-16
SLIDE 16

Plan Adaptation (cont’d)

16

!imagine ?task ?arg1 ?arg2 !imagine !move_base_param ?area slow/fast SHOP2:

grasp_object_w_arm
 mug1
 rightarm1 assume_manipulation_pose
 manipulationareaeastcounter1
 rightarm1 !pick_up_object
 mug1
 rightarm1 assume_manipulation_pose
 manipulationareaeastcounter1
 rightarm1 !move_base_blind
 premanipulationareaeastcounter1 leave_manipulation_pose
 manipulationareaeastcounter1 !move_torso
 torsoupposture

(S. Stock, Univ. Osnabrück)

slide-17
SLIDE 17

Execution & Imagination

17

  • projection: generate world state foreach !imagine

action

  • translation: converting symbolic (blackboard) values

in discrete coordinates

  • FI evaluates best confidence and shortest time
  • FI returns ordered list of parametrization + duration
slide-18
SLIDE 18
  • sampling:
  • discretization:

Sampling

18

fs = 20Hz

  • carry tall object:

z

slide-19
SLIDE 19

Handling Uncertainty

  • variables: (event 1), (event 2), ..
  • empirical coefficients: (event 1), (event 2), ..
  • confidence:

19

  • carry tall object:

c1 c2 a1 a2

slide-20
SLIDE 20

Assumptions & Limitations

  • discrete thresholds (empirically defined)
  • discrete (action) parameter set
  • one action per imagination (plan-step sync)
  • performance correlation to simulation granularity
  • exogenous events rarely occur
  • ”challenging“ simulation

20

slide-21
SLIDE 21

Experiments

  • 1. simulation validation
  • 2. recognition & manipulation
  • 3. serve a coffee
  • 4. carry a tall object

21

slide-22
SLIDE 22

Simulation Validation

22

PMA EastCounter1 PMA WestTable2 PMA EastTable2 nearStartArea1 PMA SouthTable1

counter1 table1 table2

North

PMA NorthTable1

slide-23
SLIDE 23

Simulation Validation (cont’d)

23 1 2 3 4 5 6 7 8 9 10 11 12 Duration Mean in [s] 5 10 15 20 25 30 1 2 3 4 5 6 7 8 9 10 11 12 Standard Deviation in [s] 1 2 3 4 5 Simulation PR2

1 mb_f EC1/ET2 2 mb_s EC1/ET2 3 mb_f EC1/NT1 4 mb_s EC1/NT1 5 mb_f ET2/EC1 6 mb_s ET2/EC1 7 mb_f NT1/EC1 8 mb_s NT1/EC1 9 mt U/D 10 mt D/U 11 ta U/T 12 ta T/U

slide-24
SLIDE 24

Recognition & Manipulation

24

Recognition Probability

0,49 0,51 0,54 0,56 0,58 front back left right

torso up torso down

slide-25
SLIDE 25

Serve a Coffee

25

standard prediction 75 150 225 300

execution re-planning/imagination

counter

North

robot SouthTable1 PMA NorthTable1 PMA table1

Fail

(ICRA 2014)

slide-26
SLIDE 26

Carry a Tall Object

26

(IROS 2015)

slide-27
SLIDE 27

Carry a Tall Object

27

(IROS 2015)

slow fast

slide-28
SLIDE 28

My code online available..

  • https://github.com/buzzer/pr2_imagination
  • https://github.com/buzzer/tams_pr2

28

slide-29
SLIDE 29

Conclusion

  • improved task-planning system with

”functional imagination“

  • ”common-sense“ physics-based

prediction (cf. Marques, De Silva)

  • enabling hybrid reasoning (cf. Moffit)
  • based on action parametrization (cf.

Beetz)

  • probabilistic projection (sampling,

confidence)

  • validation of simulation
  • new system on a PR2 (out-of-the-box

HTN + Gazebo)

29

slide-30
SLIDE 30

Future Research

  • improve the robot’s performance in changing and

partly unknown worlds

  • partial plan imagination
  • continuous representations (parameter + sampling)
  • exploit temporal aspects (re-scheduling)
  • reactive perception + imagination
  • cloud robotics

30

slide-31
SLIDE 31

31

Thank you for your attention!

slide-32
SLIDE 32

References

Lavindra de Silva, Amit Kumar Pandey, and Rachid Alami. An interface for interleaved symbolic-geometric planning and backtracking. In Proceedings of the International Conference on Intelligent Robots and Systems (IROS), 2013. Hugo Gravato Marques, Owen Holland, and Richard Newcombe. A modelling framework for functional imagination. In AISB Convention, pp. 51–58, 2008. Michael D. Moffitt and Martha E. Pollack. Optimal rectangle packing: A meta- CSP approach. In ICAPS, pp. 93-102. AAAI, 2006 Michael Beetz. Concurrent Reactive Plans: Anticipating and Forestalling Execution Failures. Springer-Verlag, Berlin, Heidelberg, 2000. Vere, S. A. Planning in Time: Windows and Durations for Activities and Goals. IEEE Transactions on Pattern Analysis and Machine Intelligence, PAMI- 5(3): 246-247, 1983.

32