IFAC 2004

1
Computer Science Software Engineering
Autonomously Controlled
Light Rail Vehicle
Thomas J. Byrne
Supervisor Dr. Charles Markham
2
Introduction

• Conventional Light Rail Vehicles use Human
  Perception alone to Control the Vehicle.

3
Introduction

• Conventional Light Rail Vehicles use Human
  Perception alone to Control the Vehicle.
• Introducing a Range Finder can allow the Vehicle
  Speed to be Controlled Automatically depending on
  the range of objects.
• Uses in Fully Autonomous System or Driver
  Assisted System.

4
Requirements

• Develop and Implement
• Interface with the Range Finder
• Speed Control System of the LRV
• A GUI that demonstrates the internal Operation of
  
• the system
• Path Prediction Routines based on the Kalman
  filter
• Real Time Working Model of the system

5
System Components

• Hardware Components
• Windows 2000 Machine For Operation Of Control
  System
• Sick LMS 200 LIDAR Range Finder
• Serial Communication - RS232
• Model LRV
• Parallel Communication

6
SICK LMS 200 Range Finder

• Active LIDAR Sensor
• Emits a Infrared Laser Beam
• Mirror rotating about 180 angle
• One Degree Steps - Half Degree Steps
• Range of an object at a given angle
• Distance/Range (D)
• Time of Flight (T)
• Signal Speed (Speed of Light) (S)
• D S T

7
System Components

• Software Components

8
System Components

• Software Components

9
System Components

• Software Components

10
System Components

• Software Components

11
System Components

• Software Components

12
Implementation - Threads

• Win32 API Multithreading
• Simultaneous
• Reading from Serial Port and LRV Speed Control
• Updating of GUI

13
Implementation - GUI
Plotting
Status Information

• Speed
• Direction
• Brake
• Baud Rate
• Resolution

14
Implementation Object Detection
Less than 100 mm

• Consecutive Range Points are compared to
  determine if there is a rapid change in range
  data.

Greater than 600 mm

• If two points have a difference greater than
  600mm a new object has been located.

• The left and right most positions are converted
  from polar to Cartesian coordinates and stored of
  use by the Kalman Routine.

(x,y)
(x,y)
15
Implementation Object Detection

• Euclidian Distance

• Kalman Filter

• Initial State Vector
• x position
• Velocity in the x Direction
• y position
• Velocity in the y Direction
• State Vector is Updates with each Iteration
  (includes noisy data)
• After Multiple Iterations the Filter Estimates
  the Future Position
• Implementation Requires Further Development

16
Thank You.
Questions