Katie Dellaquila Jeremy Nelson Khiem Tong. Project Overview [KED] Multidisciplinary Aspects [KED] Motivation (Similar Products) [KED] System Schematic.

Vehicle Targeting System

Katie DellaquilaJeremy NelsonKhiem Tong

Agenda

Project Overview [KED] Multidisciplinary Aspects [KED] Motivation (Similar Products) [KED] System Schematic [JSN] Components

Image Processing [KDT] Mechanical Control [KDT/JSN]

Testing [KED] Integration [KED] Project Feasibility and Anticipated Problems [JSN] Cost [JSN] Questions

Project Overview

This project will implement a game involving a moving target with a mechanical targeting and shooting system. 

The user playing the game will move the target around while the targeting system attempts to shoot it with foam darts. 

The system will use a web camera placed on a tripod and image processing algorithms to detect, track, and predict the location of the target.

Multidisciplinary Aspects

Mechanical Engineering Firing Mechanism, Pneumatics

Electrical Engineering Power considerations, Power MOSFETS

Software Engineering Software design principles, design

patterns, source control.

Motivation (Similar Products)

Modern weapon systems have very complex targeting systems that utilize lasers.

Target tracking for security cameras. Any application that requires the

tracking of a target. Filters and feedback system would change but idea stays the same.

System Schematic

Time-out System

If microcontroller loses contact with laptop after specified time, it will shutdown.

The microcontroller will not move the mechanisms nor fire any darts if it does not receive messages containing these commands from the laptop.

Circuit Diagrams

Power MOSFET or AC Relay for solenoid valve control

Image Processing

Three filters are used to isolate target and get it’s dimensions. Color filters iterate over

array and compare to threshold values

Each new frame captured from the camera generates an event whose listener applies the filters.

Detected object’s coordinates are then used in order to calculate the centroid.

HSL Filter

Grayscale Filter

BlobCounter

Filter

Coordinate

Translation

Target Predictio

n

Frame

Control signals to

μController

Operating Restrictions/Assumptions

Target is a solid, predefined color. Target is within operating area, defined at

calibration. Lighting conditions are “reasonable”. No external objects are the same color as

the target. Computation load affects image processing,

frame rate. Disable “RightLight” automatic light

adjustment for camera. Jitter needs to be tested.

GUI Prototype

Calibrate functionality allows for adjustments to environmental conditions.

Calibrate is done by placing uniquely colored objects in viewing area to form reference points which are stored.

Target Prediction

Simple Technique: Linear Interpolation The location of the target will be tracked. This information will be used to determine the

next location of the target. Complicated Technique: Kalman Filter

This technique will probably be more accurate, but it is more costly in terms of performance.

Timing Considerations The calculations must be done quickly for

better accuracy. Dart flight time is taken into account

Mechanical Firing Mechanism

Based on, with permission, mechanism utilized in the Remote Control Turret by Josh Bookout et al.

Barrel from Nerf gun. Machine mounting with

two holes. Servo rotates barrel

using shaft connected by 2:1 gear ratio for 360˚ rotation.

Barrel is aligned with compressed air tube.

CAD Drawings

[Source: Remote Controlled Turret]

Mechanical Specifications

Servo is rated for 2.75 lbs (4.8V) and 0.19 s/60˚

The firing rate will be limited by a minimum wait time between shots.

Range is limited by pressure which controlled by regulator.

This is not huge concern since range of accurate image processing is only 15-20’.

Testing

Unit tests of individual components. Mechanical and software tests separately Mechanical▪ Empirical tests to produce scatter plot of dart firing

locations▪ Timing for updating servo position, barrel rotation, and

travel time. C# Code components▪ Image processing (light conditions, camera location, etc.)▪ Target prediction (accuracy, timing, etc)

Embedded Code components▪ Serial communication with laptop▪ Servo control, solenoid valve control

Integration

SVN server setup to help with software integration (and testing).

Iterative integration. Laptop and microcontroller

communication Microcontroller and mechanical

components control Overall system – accuracy and usability

Feasibility/Projected Problems

Mechanical design components Borrowed CAD drawings have very

precise dimensions, which make machining parts time costly

Lighting conditions Target prediction/calibration will be

challenging.

Cost

Component Retail Price Our Price

Webcam *$70 $0

Compressed Air Tank *$40 $40

Foam Dart Launcher $30 $30

Solenoid Valves **$40 $0

Turret Platform *$50 $50

Microcontroller *$100 $0

Ultrasound Sensor $15 $0

Vehicle $10 $0

Laptop $1200 $0

Total $1555 $120

*Purchased by RIT Department of Computer Engineering** Received from Mr. Wellin

DEMO

Image processing GUI Target Detection

Acknowledgements

1. Remote Controlled Turret, Josh Bookout et al. [Spring 2008]

Questions

Any questions or comments?