1. Robot Laser Tag Team 1: Matt Pruitt, Dirk Van Bruggen, and Jay Doyle
Team 7: Jason Grant & Joe Thompson
Artists: Elizabeth Maltby, Kayla Wolter, and Chelsea Young
Dirk

2. Overview Project Overview Methodology / Approach Results
Collaboration
Team 1
Team 7
Results
Demonstration Video
Discussion & Future Work
Questions
Dirk

3. Project
Overview
Matt

4. Why Laser Tag? Involves moving and thinking about moving
Fits with Turtlebot constraints
Made use of available sensors on the robots
Vision and 3D sensors with cliff/bump as backup
Shooting easily simulated with vision system
Fun for kids to watch and to play
Can also be controlled by PS3 controller
Conceptually easy game
Matt

5. Goals
To have multiple robots autonomously play a game of laser tag in a constrained environment
Move through the environment quickly while avoiding obstacles
Track another robot when it is discovered
"Shoot" another robot when it is close enough
Scatter when it is hit
Develop independent algorithms for autonomous gameplay
Matt

6. Methodology
Joe

7. System Overview Central game manager Scoreboard
Handles score
Hit/miss decided by neutral party
Controls game state (hunting, scatter)
ROS Master Node
Scoreboard
Scores maintained by MySQL
PHP website to display score on projector
Two independent players
Each Turtlebot has own control algorithm
Communicate with ROS master over WiFi
Joe

8. Central Game Manager (ROS Master)
Handles Score
Hit Decision
Uses CMVision blob info to determine hit/miss
Had capability to process blobs without CMVision
Switched to processing blobs on robots due to network latency
Controls game state (hunting, scatter)
Custom ROS msgs to send game state to players
Notifies of state change & hit/miss
ROS Master Node
Joe

9. Scoreboard MySQL database for holding score
PHP script for displaying score on webpage
Apache web server for hosting score page
Matt

10. Independent Players Each player has own strategy
Standard msgs agreed upon for game state
Custom ROS msgs for game state were created
Each player can be controlled through PS3
Controller paired with robot and used PS3 teleop
Robot would be autonomous until controller button pressed
Jason

11. Team 1 Outline Component Diagram State Machine PCL Processing
PS3 Controller
Jay

12. Team 1 - System Architecture
Game Arbitration Server
Bump/Cliff Sensor
Central Navigation Thread
Jay
CMVision Blobs
Kinect/PCL Clouds
ROS Messages

13. Team 1 - Navigation
Jay

14. Team 1 - Navigation
Jay

15. Crazy
Ivan
Jay

16. Team 1 - Controller
Matt

17. Team 7 - Outline Separation of control and perception State Management
Robot Control
Issues / Solutions
Joe

18. Team 7 - Perception
All sensing through the Kinect handled by a nodelet
Provides speed for rapid PCL processing
Handles blob processing from CMVision
Allows for translation from noisy sensor data to higher level information before reaching control
Environment
Perception
Obstacles
Joe
PCL
Depth
Collision
Opponent
CMVision
RGB
Target

19. Team 7 - Custom Messages Opponent Obstacles Target Collision
Contains : Blob information for opponent's "statue" aligned with PCL data
Purpose : Provides information on position of opponent
Obstacles
Contains : Centroid information for PC points between 1 and 3 meters away within a half a meter to the right or left
Purpose : Provides long range information of field
Target
Contains : Blob information for opponent's target aligned with PCL data
Purpose : Provides information for a shooting solution
Collision
Contains : Centroid information for PC points < 1 meter away
Purpose : Provides information concerning imminent collision
Jason

20. Team 7 - State Machine
Jason

21. Team 7 - Controller Reactive Controller
Default Movement : move forward
Each message sparks a reaction function
Each assign weights
An arbitrator calculated a weighted average and published a final rotational and linear velocity to the TurtleBot
Allows larger goals to be usurped by immediate goals
Larger Goals : Hunt Opponent, Hide by Obstacles
Immediate Goals : Avoid Collisions
Joe

22. Team 7 - Controller Movement Arbitrator Move Avoid Collisions
HUGE
Hunt Opponent
WEIGHT
Opponent
Medium Weight
Hide by Obstacles
Movement Arbitrator
Obstacles
Small Weight
Joe
In range?
Move
Target
Yes
Shoot

23. Jason

24. Unexpected Events
Formed line for each team for unexpectedly large crowd
Wi-Fi Range
Far side of area was not covered by Wi-Fi
Running into walls
Team7: default behavior was go forward
Team1: kinect did not see paper at small angles
Running through wall
Avatars could rotate on robot base, kids confused
Changing arenas
Battery life
One robot was a ballerina
Jason

25. Problems with Multi-Robot System
Passing Messages
Multiple Namespaces to Avoid Conflicting Msgs
Wi-Fi interference in Joyce
Multiple Namespaces
All default Turtlebot scripts changed
e.g. both robots cannot publish/subscribe to cmd_vel
CMVision needed to be remapped
Nodelets difficult to integrate with new namespace
No existing tutorials covered topic completely
Dirk

26. Demo Video
Jay

27. Jay!

28. Jay

29. Jay

30. Discussion
Jason

31. Discussion Robots played laser tag Add "friggin" lasers 2013 Students:
Dig into ROS early in semester
Have lab assignment using multiple robots
Turning issues with low battery
Jason

32. Questions
Dirk

33. Questions
Dirk

34. Questions

35. Questions

36. Video /