Top Banner
1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008
45

1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

Dec 19, 2015

Download

Documents

Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

1

Single Robot Motion Planning - II

Liang-Jun ZhangCOMP790-058

Sep 24, 2008

Page 2: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

2

Review: C-space

Workspace Configuration Space

xyRobot

Initial

Goal

Free

Obstacle C-obstacle

A 2D Translating Robot

Page 3: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

3

Review: Computing C-obstacle

• Difficult due to geometric and space complexity• Practical solutions are only available for

– Translating rigid robots: Minkowski sum– Robots with no more than 3 DOFs

Page 4: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

4

Outline

• Approximate cell decomposition

• Sampling-based motion planning

Page 5: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

5

Approximate Cell Decomposition (ACD)

• Not compute the free space exactly at once• But compute it incrementally

• Relatively easy to implement– [Lozano-Pérez 83]– [Zhu et al. 91]– [Latombe 91]– [Zhang et al. 06]

Octree decomposition

Page 6: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

6

full mixed

empty

Approximate Cell Decomposition

• Full cell

• Empty cell

• Mixed cell– Mixed– Uncertain

• Cell labelling algorithms– [Zhang et al 06]

Configuration Space

Page 7: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

7

Finding a Path by ACD

Goal

Initial

Page 8: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

8

Connectivity GraphGf : Free Connectivity Graph G: Connectivity Graph

Gf is a subgraph of G

Page 9: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

9

Finding a Path by ACD

Goal

Initial Gf : Free Connectivity Graph

Page 10: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

10

Finding a Path by ACD

L: Guiding Path• First Graph Cut Algorithm– Guiding path in connectivity

graph G

– Only subdivide along this path

– Update the graphs G and Gf

Page 11: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

11

First Graph Cut Algorithm

Only subdivide the cells along L

L : Guiding Path

new Gf

Page 12: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

12

Finding a Path by ACDGf

• A channel

• Can be used for path smoothing.

Page 13: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

13

ACD for Path Non-existence

C-space

Goal

Initial

Page 14: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

14

Connectivity Graph Guiding Path

ACD for Path Non-existence

Page 15: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

15

ACD for Path Non-existence

Connectivity graph is not connected

No path!

A sufficient condition for deciding path non-existence

Page 16: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

16

• Live Demo– Gear-2DOF– Gear-3DOF

Page 17: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

17

Five-gear Example

Video

Initial

Goal

roadmap in free space

Total timing 85s

# of total cells 168K

Page 18: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

18

Two-gear Example

no path!

Cells in C-obstacle

Initial

Goal

Roadmap in F

Video 3.356s

Page 19: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

19

Motion Planning Framework

• Continuous representation

• Discretization

• Graph search

Page 20: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

20

Summary: Approximate Cell Decomposition

• Simple and easy to implement

• Efficient and practical for low DOF robots – Inefficient for 5 or more DOFs robot

• Resolution-complete– Find a path if there is one– Otherwise, report path non-existence– Up to some resolution of the cell

Page 21: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

21

Outline

• Approximate cell decomposition

• Sampling-based motion planning

Page 22: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

22

Motivation

• Geometric complexity• Space dimensionality

Page 23: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

23

Probabilistic Roadmap (PRM)

free space

qqinitinit

qqgoalgoal

milestone

[Kavraki, Svetska, Latombe,Overmars, 95]

local path

Page 24: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

24

Basic PRM AalgorithmInput: geometry of the moving object & obstaclesOutput: roadmap G = (V, E)

1: V and E .

2: repeat3: q a configuration sampled uniformly at random from C.4: if CLEAR(q)then5: Add q to V.6: Nq a set of nodes in V that are close to q.

6: for each q’ Nq, in order of increasing d(q,q’)7: if LINK(q’,q)then8: Add an edge between q and q’ to E.

Page 25: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

25

Two Geometric Primitives in C-space

• CLEAR(q)Is configuration q collision free or not?

• LINK(q, q’)

Is the straight-line path between q and q’ collision-free?

Page 26: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

26

Query Processing

• Connect qinit and qgoal to the roadmap

• Start at qinit and qgoal, perform a random walk, and try to connect with one of the milestones nearby

• Try multiple times

Page 27: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

27

Two Tenets of PRM Planning Checking sampled configurations and

connections between samples for collision can be done efficiently. Hierarchical collision checking[Hierarchical collision checking methods were developed independently from PRM, roughly at the same time]

A relatively small number of milestones and local paths are sufficient to capture the connectivity of the free space. Exponential convergence in expansive

free space (probabilistic completeness)

Page 28: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

28

Why does it work? Intuition• A small number of milestones almost

“cover” the entire free space.

Page 29: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

29

Two Tenets of PRM Planning Checking sampled configurations and

connections between samples for collision can be done efficiently. Hierarchical collision checking[Hierarchical collision checking methods were developed independently from PRM, roughly at the same time]

A relatively small number of milestones and local paths are sufficient to capture the connectivity of the free space. Exponential convergence in expansive

free space (probabilistic completeness)

Page 30: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

30

Narrow Passage Problem

• Narrow passages are difficult to be sampled due to their small volumes in C-space

Narrow passage

Alpha puzzle

qinitqgoal

2F

3F

Configuration Space

1F

Page 31: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

31

Difficulty• Many small connected components

Page 32: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

32

Strategies to Improve PRM

• Where to sample new milestones?– Sampling strategy

• Which milestones to connect?– Connection strategy

• Goal: – Minimize roadmap size to correctly answer

motion-planning queries

Page 33: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

33

Sampling Strategies

Page 34: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

34

small visibility sets

good visibility

poor visibility

Poor Visibility in Narrow Passages

• Non-uniform sampling strategies

Page 35: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

35

But how to identify poor visibility regions?

• What is the source of information?– Robot and environment geometry

• How to exploit it?– Workspace-guided strategies– Dilation-based strategies– Filtering strategies– Adaptive strategies

Page 36: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

36

Identify narrow passages in the workspace and map them into the configuration space

Workspace-guided strategies

Page 37: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

37

F

O

Dilation-based strategies • During roadmap construction, allow milestones

with small penetration• Dilate the free space

– [Hsu et al. 98, Saha et al. 05, Cheng et al. 06, Zhang et al. 07]

A milestone with small penetration

Page 38: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

38

Error

• If a path is returned, the answer is always correct.

• If no path is found, the answer may or may not be correct. We hope it is correct with high probability.

Page 39: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

39

Weaker Completeness

Complete planner Too slow Heuristic planner Too unreliable

Probabilistic completeness:If a solution path exists, then the probability that the planner will find one is a fast growing function that goes to 1 as the running time increases

Page 40: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

40

Kinodynamic Planning

Page 41: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

41

RRT for Kinodynamic Systems

• Rapidly-exploring Random Tree

• Randomly select a control input

Page 42: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

42

More Examples

• Car pulling trailers (complicated kinematics -- no dynamics)

Page 43: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

43

Summary: Sampling-based Motion Planning

+ Efficient in practice + Work for robots with many DOF (high-

dimensional configuration spaces)+ Has been applied for various motion planning

problems (non-holonomic, kinodynamic planning etc.)

- Narrow passages problems (one of the hot areas)

- May not terminate when no path exists

Page 44: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

44

Summary

• Configuration space

• Visibility graph

• Approximate cell decomposition – Decompose the free space into simple cells and represent the

connectivity of the free space by the adjacency graph of these cells

• Sampling-based approach– High-dimensional Configuration Spaces– Capture the connectivity of the free space by sampling

Page 45: 1 Single Robot Motion Planning - II Liang-Jun Zhang COMP790-058 Sep 24, 2008.

45

References

• Books– J.C. Latombe. Robot Motion Planning, 1991.– S.M. LaValle, Planning Algorithms, 2006

• Free book: http://msl.cs.uiuc.edu/planning/

– H. Choset et al. Principles of Robot Motion: Theory, Algorithms, and Implementations, 2005

• Conferences– ICRA: IEEE International Conference on Robotics and

Automation– IROS: IEEE/RSJ International Conference on Intelligent RObots

and Systems– WAFR: Workshop on the Algorithmic Foundations of Robotics