You are on page 1of 31

May 2012 1

Motion Planning
Shmuel Wimer
Bar Ilan Univ., Eng. Faculty
Technion, EE Faculty

May 2012 2
Outline
Problem definition
Point robot
Work space and configuration space
Minkowski sums
Translational motion planning
Rotational motion planning
May 2012 3
Types of Robots and Motions
Articulated Robot
Translation
Motion
start
goal
May 2012 4
Rotational
Motion
start
goal
Some robots can move in any direction (e.g., ants)
Some robots cannot translate (e.g., cars)
Well study translational and rotational motions
Given a robot, is there a free paths (no collisions) from start to goal?
May 2012 5
Work Space and Configuration Space
Reference
Point
2 Degrees of freedom 3 Degrees of freedom
May 2012 6
Configuration space
Work space
May 2012 7
Free Space Computation
Divide into trapezoids. It
takes O(nlogn) expected time.
Remove trapezoids of
obstacle in O(n) time.
May 2012 8
Building a Road Map
Allocate node at center
of vertical edges
Allocate node at center
of every trapezoid
Connect center nodes
to edge nodes
Done in O(n) with doubly-connected edge list
May 2012 9
Computing a Path
Get from start to center
of trapezoid in O(logn)
time
Get from goal to center
of trapezoid in O(logn)
time
Connect center of
trapezoids by BFS in
O(n) time
May 2012 10
May 2012 11
Convert a problem with polygonal robot into point robot by modifying the
obstacles in the configuration space to incorporate the geometry of the
robot.
Minkowski Sums
obstacle
robot
May 2012 12
May 2012 13
May 2012 14

May 2012 15
Extreme Points and Directions
May 2012 16
How complex is Minkowski sum of two polygons?
May 2012 17

May 2012 18
-
R
a
b
c
d
-
R
a
b
c
d
-
R
a
b
c
d
-
R
a
b
c
d
-
R
a
b
c
d
-
R
a
b
c
d
Construction of Minkowski Sum
1
2
3
4
5
P
Edges of P and R are labeled
counterclockwise
b
1
2
3
4
5
c
d
d
c
a
a
b
May 2012 19
1
2
3
4
5
a
b
c
d
1
2
3
4
5
P
-
R
a
b
c
d
May 2012 20
b
1
2
3
4
5
c
d
d
c
a
a
b
1
2
3
4
5
a
b
c
d
May 2012 21
May 2012 22
Rotation Moving a Ladder
Minkowski sum for ladder
at 0 rotation. Blockage
exists.
May 2012 23
Minkowski sum for ladder at
30 rotation. Blockage exists.
Minkowski sum for ladder at
60 rotation. Blockage
changed.
May 2012 24
Conversion to 3D Motion Problem
Bottom view
Front view. varies from 0 to
75.
Ladders reference point can move in the 3D space!
May 2012 25
Cell Decomposition
Minkowski sums
A
B
C
Cell decomposition
1
0
8
9
2
1
3
4
5
6
7
Obstacles

R
Ladder
May 2012 26
A Cell is the collection of all free points labeled
with the same front/back edge label pairs.
A: (3,2); B: (3,8); C: (1,9)
Cell decomposition has discontinuities when
ladder is oriented similar to an edge.
There are finite number of ladder rotation where
cell decomposition is changing.
New cells can appear and old ones may disappear.
May 2012 27
5
6
7
1
0
8
9
2
1
3
4

R
A
B
C disappeared
Ladder is rotated
May 2012 28
In Connectivity Graph G nodes are cells of decomposition and edges are
connecting nodes corresponding to adjacent cells in free area (a kind of dual
graph).
A: (3,2)
B: (3,8)
(1,8)
C: (1,9)
(1,)
(10,)
(5,)
(3,)
G0
May 2012 29
(4,)
(1,)
(10,)
(5,)
(3,)
A: (3,2)
B:
(3,8)
(1,8)
(7,8)
Critical Orientations correspond to slants of edges.
May 2012 30
Connectivity Graph G is constructed by stacking the connectivity
graphs G corresponding to the critical orientations.
Vertices of two distinct G are connected iff they are labeled with the
same edge pair. Starting from G0, G are added in increasing order of ,
thus creating a layered 3D graph.
A paths from start to goal if exists can be found by a BFS algorithm.
May 2012 31
History

You might also like