LamEGO

1
LamE-GO

• a Lego RCX 2.0 Project
• with Network Remote Controlling
• Video Streaming and Image Processing
• Point-2-Go Navigation

2
Project Team Members

• Finland
• Teppo Kurki (Team Leader)
• Antti Levo
• Sweden
• Daniel Holm
• Fredrik Höglin
• Björn Magnusson
• Joel Samuelsson

3
Problem Statement Refinement

• Client Server Communications
• Provides means to send control messages between
  server and client
• Server Application
• Communicates with client and robot
• Controls video streaming, image processing and
  navigation
• Video Streaming
• RTP Transmitter at the server and receiver at the
  client
• An ability to take screenshots
• Navigation Point-2-Go -interface
• Image processing
• Detect obstacles from a still image and generate
  a map
• Find the robots location and orientation
• Actual navigation
• Receive target coordinates from the client
• Calculate a path and control the robot to reach
  the target coordinates

4
Problem Statement Refinement

• Client Application
• A single window. Shows all control buttons and
  video stream
• Server Robot Communications
• Allows the robot and the server to send messages
  to each other over IR-link
• Robot Software
• Controls the robots movement according to
  received messages
• Reacts to sensor inputs
• Two touch sensors to detect collisions front
  and rear
• Light sensor to prevent dropping off a table

5
Milestone Meeting 1 (Teppo)

• First official IRC meeting on 25.1.
• Team members got to know each other
• Team leader was selected
• Persons responsible for Milestone meetings and
  reports were selected

• An attempt to prepare the SVN for use was
  unsuccessful
• First rough draft of the high-level
• architecture was made
• First development decisions
• Java for client and server applications
• RCX 2.0 kit for the robot
• Lejos for the robot software

6
Milestone Meeting 2 (Björn, Joel)

• Progress
• Design Document
• Working groups

7
Milestone Meeting 2 - Design

• Client -gt Server communication
• Server Application
• Video Streaming
• Client GUI
• Server -gt Robot communication
• Robot
• Robot AI
• Excellence areas

8
Milestone Meeting 3 (Antti, Teppo)
9
Milestone Meeting 3 - ?
10
Milestone Meeting 4 (Daniel, Fredrik)

• Setbacks
• Robot Communications
• Point-2-Go
• Video Streaming
• Achievements
• Redesign
• Integration
• Testing

11
High-level System Design
12
Division of labor

• Integration done in co-operation with the
  applicable subteams

13
Details on Design, Version Control, Documentation
and Packaging

• High-level design and communication interfaces
  were agreed on among all members of the group
• More detailed design done by each subteam
  independently
• Version control using SVN
• If there was a need to modify other subteams
  code, the corresponding subteam was consulted
  before committing changes
• Elaborate documentation to each file using
  JavaDoc the actual documentation is easy to
  generate
• No actual software packages were created only
  three sets of files

Client
Robot
Server
14
Client Application
15
Client GUI

• Working as intended
• Biggest problem
• Desired functionality

16
Robot - Server Communications Current Version
17
Robot Server Communications - Previous Version

• No Concurrency - RCX locks up
• Threads - Many problems.
• No timeout on messages
• Other problems (compiling, testing)

18
Robot AI
19
Robot AI - Progress

• Success
• Problems
• Solutions

20
Server Application
21
Client Server Communications

• First ready part of the project!
• Communicating classes implemented as threads
• Simple design using java.net.Socket
• No notable problems in implementation

22
Point-2-Go Navigation

• Our objective was to allow user to click on the
  video at the client end, and then the server-side
  software would simply control the robot to reach
  the clicked point, while avoiding obstacles
• Consists of two parts Image Processing and Path
  Calculation
• Path Calculation also includes the robot
  controlling logic
• Image Processing
• ImageProcessor gets as an input a still image
  from the VideoHandler
• Generates a map from the image with convolution
  and thresholding
• Finds the robot which is equipped with green and
  blue pieces in the front and back, respectively
• Determines the robots direction
• Offers this information to be used by Path
  Calculation
• Image Processing parts are fully implemented and
  functional, although results may be susceptible
  to lighting of the environment
• For reasonable results, the operating area should
  be free of other green- or blue-coloured objects

23
Point-2-Go Navigation

• Path Calculation

24
Video Streaming Slide 1
25
Video Streaming Slide 2
26
The End

• Questions ?
• Comments ?
• -
• Feel free to provide both !