Mobile Robot Applications

1
(No Transcript)
2

• Mobile Robot Applications
• Textbook
• T. Bräunl Embedded Robotics, Springer 2003
• Recommended Reading
• 1. J. Jones, A. Flynn Mobile Robots, 2nd Ed., AK
  Peters, 1999
• ? Hobbyists introduction, easy reading
• 2. R. Arkin Behavior-based Robotics,
• ? Overview of behavior-based robotics
• 3. Kernighan, Ritchie The C Programming Language
• alternatively ltany C programming bookgt
• ? C programming skills are important!

3

• Contents Topics
• Maze driving
• Micro Mouse Contest
• Mapping
• Driving in unknown environments
• Elementary Image Processing
• Edge detection, color detection, color blobs
• Robot Soccer
• autonomous agents

4
Mazes and Mapping
robot
Know where to go!
Place p
Explore while finding the connection.
5
Mazes
We won local competition in 1990 Two our teams
did not complete the run 2004
6
(No Transcript)
7

• This is becoming a competition for sensors,
  motors and crazy ideas.
• Algorithmic problems are already solved.

8
Long rods for sensing
9
Cell-based maze for mapping and motion planning
10
In early contests you can win using this simple
algorithm. Next it was changed to make contest
more interesting

• This will not find the object in the middle if
  there is much empty space around.

11
Follow left wall Algorithm
Explore_left Many Probabilistic variants have
been created
x,y coordinates, dir direction
flags
See next page for these routines
12
Depending on current direction, update x and y
coordinates of the mouse
13

• Never finds the gold
• Idea to remember there are good special
  algorithms for some kinds of mazes.
• If you deal with general space or irregular map
  of labyrinth, you have to use several algorithms
  and adapt.

There are many recursive algorithms, we will
illustrate one of them
14
Left wall following
recursion
15
In backtrack point robot knows that it has done a
bad decision

• This explains and illustrates the concept of
  backtracking that is fundamental to robotics and
  AI

16
Explore will call itself recursively
Mark x and y position
Check situations if front open etc Set flags
front open etc
Use flags front open etc
17
Recursive call of itself
This part shows recursive calls in all situations
Front open, Left open and right open
18
We can combine recursion and left -wall-following
algorithms in several ways
19

1. Discuss how it works.
2. How it is represented.

• This map shows calculating distances from the
  start for labyrinth from bottom left

Using grid we start from here and go everywhere
adding 1 at each step
One approach to solve this are the Flood Fill
Algorithms
20
Flood Fill Algorithms

• The idea of marking cells appears here again

21
Algorithm continued
22
continuation
Example on next slide
23
Phase 3
Phase 2
Phase 1
This is like breadth first search
24
Next Stage of Flood Algorithm Shortest Path

• Now we have
• Explored the maze
• Know the distance to goal from every cell
• Missing
• Shortest path from start to goal
• Idea
• Generate shortest path from goal backward to
  start

25
(No Transcript)
26
(No Transcript)
27
What to visualize in maze algorithms

• Distances of cells from start position

• Part of map that has been covered so far

• Map of labyrinth

• Path already done by robot

28
Real-world mazes (hospitals, universities) and
labyrinths (forest, park, open battlefield)

• Applications in hospitals, museums, mines, big
  government buildings.

Learn from counting doors or information on walls
29
Mapping

• Mapping an unknown environment is similar to the
  maze problem
• However, maze is very simple
• fixed size cells
• only 90º angles
• Now let us look at general environments

30
Mapping

• Explore unknown environment
• Use infra-red PSD and infra-red proxy sensors
  only
• Apply DistBug algorithm for wall following once
  an obstacle is encountered
• Enter sensor measurement data in map
• Use visibility graph with configuration space
  representation

31

• Exploring cells of the map grid based

32
continued
Exploring obstacles in the map - general maps,
shapes, no grid.
33
This slide explains how to use grids to draw the
map based on sensor information and actions
executed.
34
This slide explains how to use grids to draw the
map based on sensor information and actions
executed.

• Such parts can be next fixed based on general
  predetermined knowledge of the nature of walls,
  obstacles and sizes.

35
The smaller the error the more accurate the map
36
(No Transcript)
37
(No Transcript)
38
You should collect these kinds of data for your
robot environment of the demo. Think in advance
where our robots will be demonstrated. Deans
attrium? Near elevators? Not the lab!!
39
Conclusion

1. Now that you understand one application of
   search, go read again the slides about search
   algorithms and think how they can be used in this
   application.
2. What can be the cost (fitness) functions?
3. Think about other mapping algorithms. Can you use
   randomness?