34
Fall 2005 1 Path Planning from text

Path Planning from text

  • Upload
    duke

  • View
    45

  • Download
    1

Embed Size (px)

DESCRIPTION

Path Planning from text. Outline. Point Robot Translational Robot Rotational Robot. Visibility Graph (Point Robot). goal. Edges between all pairs of visible vertices. start. Use graph algorithm to find a path from start to goal. Free Space (Point Robot). Path Planning (Point Robot). - PowerPoint PPT Presentation

Citation preview

Page 1: Path Planning from text

Fall 2005 1

Path Planning from text

Page 2: Path Planning from text

Fall 2005 2

Outline

Point RobotTranslational RobotRotational Robot

Page 3: Path Planning from text

Fall 2005 3

Visibility Graph (Point Robot)

start

goal

Edges between all

pairs of visible vertices

Use graph algorithm to find a path from start to goal

Page 4: Path Planning from text

Fall 2005 4

Free Space (Point Robot)

Page 5: Path Planning from text

Fall 2005 5

Path Planning (Point Robot)

Page 6: Path Planning from text

Fall 2005 6

Path Planning (cont)

Page 7: Path Planning from text

Fall 2005 7

Robot (translational)

polygonal

Page 8: Path Planning from text

Fall 2005 8

C-space Obstacle of P

Page 9: Path Planning from text

Fall 2005 9

Minkowski Sum Coordinate

dependent!

Page 10: Path Planning from text

Fall 2005 10

Theorem

CP is P(-R(0,0))

R(0,0)

–R(0,0)

Proof: R(x,y) intersect P (x,y)P(-R(0,0))

Page 11: Path Planning from text

Fall 2005 11

R(0,0)

–R(0,0)

PqyxRq

q

),(

onintersectiin point A

)0,0(, Ryqxq yx

)0,0(, Ryqxq yx

(x,y)

If intersect, (x,y) is in CPIf intersect, (x,y) is in CP

))0,0((,

,,

RPyx

yqxqqq

Pq

yxyx

q

Page 12: Path Planning from text

Fall 2005 12

R(0,0)

–R(0,0)

))0,0((, RPyx

(x,y)

If (x,y) is in CP, R(x,y)&P intersectIf (x,y) is in CP, R(x,y)&P intersect

p

yy

xx

ryp

rxp

is,That

r

yyxx

yxyx

rprpyx

PppRrr

,,such that

, )0,0(,

Page 13: Path Planning from text

Fall 2005 13

Computing Minkowski Sum

Page 14: Path Planning from text

Fall 2005 14

Example

v1,v4

v2

v3

w1,w5 w2

w3

w4

[i,j] = (1,1)

Add v1+w1

angle(v1v2) > angle(w1w2)

j2

Page 15: Path Planning from text

Fall 2005 15v1v2

v3

w1 w2

w3

w4

[i,j] = (1,2)

Add v1+w2

angle(v1v2) < angle(w2w3)

i2

Page 16: Path Planning from text

Fall 2005 16v1v2

v3

w1 w2

w3

w4

[i,j] = (2,2)

Add v2+w2

angle(v2v3) > angle(w2w3)

j3

Page 17: Path Planning from text

Fall 2005 17v1v2

v3

w1 w2

w3

w4

[i,j] = (2,3)

Add v2+w3

angle(v2v3) < angle(w3w4)

i3

Page 18: Path Planning from text

Fall 2005 18v4,v1

v2

v3

w1 w2

w3

w4

[i,j] = (3,3)

Add v3+w3

angle(v3v4) > angle(w3w4)

j4

Page 19: Path Planning from text

Fall 2005 19v4,v1

v2

v3

w5,w1 w2

w3

w4

[i,j] = (3,4)

Add v3+w4

angle(v3v4) < angle(w4w5)

i4

Page 20: Path Planning from text

Fall 2005 20v4,v1

v2

v3

w5,w1 w2

w3

w4

[i,j] = (4,4)

Add v4+w4

Page 21: Path Planning from text

Fall 2005 21

Non-convex polygons

Page 22: Path Planning from text

Fall 2005 22

Time Complexity

It is O(n+m) if both polygons are convex.It is O(nm) if one of the polygons is convex and one is non-convex.It is O(n2m2) if both polygons are non-convex.

Page 23: Path Planning from text

Fall 2005 23

Example 2

v4,v1

v2

v3

w5,w1

w2

w3

w4

[i,j] = (1,1)

Add v1+w1

angle(v1v2) < angle(w1w2)

i2

Page 24: Path Planning from text

Fall 2005 24v4,v1

v2

v3

w5,w1

w2

w3

w4

[i,j] = (2,1)

Add v2+w1

angle(v2v3) > angle(w1w2)

j2

Page 25: Path Planning from text

Fall 2005 25v4,v1

v2

v3

w5,w1

w2

w3

w4

[i,j] = (2,2)

Add v2+w2

angle(v2v3) > angle(w2w3)

j3

Page 26: Path Planning from text

Fall 2005 26v4,v1

v2

v3

w5,w1

w2

w3

w4

[i,j] = (2,3)

Add v2+w3

angle(v2v3) < angle(w3w4)

i3

Page 27: Path Planning from text

Fall 2005 27v4,v1

v2

v3

w5,w1

w2

w3

w4

[i,j] = (3,3)

Add v3+w3

angle(v3v4) > angle(w3w4)

j4

Page 28: Path Planning from text

Fall 2005 28v4,v1

v2

v3

w5,w1

w2

w3

w4

[i,j] = (3,4)

Add v3+w4

angle(v3v4) < angle(w4w5)

i4

Page 29: Path Planning from text

Fall 2005 29v4,v1

v2

v3

w5,w1

w2

w3

w4

[i,j] = (4,4)

Add v4+w4

Page 30: Path Planning from text

Fall 2005 30

Rotational Robot

R (x, y, Ф)Ф: rotated anti-clockwise through an angle Ф

Page 31: Path Planning from text

Fall 2005 31

Rotatonal Robot Motion Plan

Piano mover applet

Page 32: Path Planning from text

Fall 2005 32

C-space of Rotational Robot

Page 33: Path Planning from text

Fall 2005 33

Path Planning (Rotational Robot)

Each slice: R(0,0,i): obtain a roadmapProject all roadmap to get “intersection” – a pure rotation from i to j

Use a slight larger robot to ensure pure rotation won’t collide with obstacles

Page 34: Path Planning from text

Fall 2005 34

Homework

P

R

[use the grid line to compute the result as accurate as possible]

1. Compute CP w.r.t. R2. Compute CP w.r.t. R’3. R and R’ are exactly the same

robot, differ only in reference point. Are CPs in 1 and 2 the same?

4. Do 1 and 2 obtain the same answer regarding to the intersection query? That is, the configuration shown left is reported as intersection in 1 & 2.

R’