Microsoft Kinect Sensor Skeleton Tracking

Published Date: 02 Nov 2017

BY

NG HUEY MUN

TARC LOGO

DIVISION OF MECHANICAL ENGINEERING

SCHOOL OF TECHNOLOGY

TUNKU ABDUL RAHMAN COLLEGE

KUALA LUMPUR

2012/2013

Microsoft Kinect Sensor skeleton tracking integrate with Roboactor puppet.

by

Ng Huey Mun

Supervisor: Engr. Eu Kok Seng, Grad. IEM

AdvDipTech ( TARC ), BEng ( Hons ),

MSc ( Sheff. Hallam )

Project dissertation submitted in partial fulfilment of the requirements for the award of Advanced Diploma in Engineering (Mechatronics Engineering)

Division of Mechanical Engineering

School of Technology

Tunku Abdul Rahman College

Kuala Lumpur

2011/2012

Copyright © 2012 by Tunku Abdul Rahman College.

All rights reserved. No part of this dissertation may be reproduced, stored in a retrieval system, or transmitted in any form or by any means without the prior permission of Tunku Abdul Rahman College.

DECLARATION

"This project submitted here with is a result of my own investigations. All information that had been obtained from other sources had been fully acknowledged. I understand that plagiarism constitutes a breach of College rules and regulations and would be subjected to disciplinary actions"

_________________

(NG HUEY MUN)

ACKNOWLEDGEMENT

I would take this opportunity to express my gratitude towards certain respectable individuals for their consistent guidance, assistance and support of doing this project. Without their guidance and support, I would not done it successfully because it was a hard way and taken a long time to complete my project.

First of all, I would like to express my sincere appreciation to my project supervisor which also my lecturer, Engr. Eu Kok Seng, who spend his own precious time and efforts in supervising this project. With the help and advice that given by him, I able to learn a lot of extra knowledge which can't find in the textbooks. Through this experience, I had improved my skill and knowledge in electronics engineering.

Besides that, I also would like to thank to my friend, Chow Seong Lip and Chiang Hung Ching for helping me in accomplishing this project in reconstructing the Roboactor robot.

Lastly, I also would like to thank to my parents who had support me in financially and mentally support. Not to forget as well for their ending support, encouragement and educate me from Tunku Abdul Rahman College.

ABSTRACT

The title of this project is Microsoft kinect sensor skeleton tracking integrate with Roboactor puppet. This project is inspired by a movie called Real Steel where a robot is being controlled by shadow movement which means the human body is the controller of the robot. The robot will follow the act and movement of the controller which is the human’s body.

To allow the robot to mimic a human body movement, the robot needs to see and processes the image of a human. Kinect sensor has the ability to detect a human body using skeleton tracking method. In Malaysia, Microsoft Kinect sensor alone is not a very popular device because it normally comes with Microsoft Xbox 360 console. Microsoft Xbox 360 console is for gaming purposes which allow gamers to experience gaming and entertainment without using a controller.

Kinect sensor detects whole human body joints and creates a skeleton image. This allows the computer to do image processing and to be able to calculate the angles between each limb and the length of each limb on the human. This information will then be transfer into Arduino microcontroller to control the servo motors inside the robot that the human wanted to control.

The purpose of this project is

TABLE OF CONTENTS

LIST OF FIGURE

LIST OF TABLE

CHAPTER 1 INTRODUCTION

Introduction

The world of engineering and technology keep changing, improving, creating and develops new innovation in various field as each new day goes by to improve humans living environment. To develop machines or devices that have the ability to communicate with its user and the environment, sensors are needed.

In the robotics world, sensors are very important. The most common types of sensors used in this robotics world are camera, infrared sensor and ultrasonic sensor. All these sensors have their own limitations. Cameras are used for near range object detection because these cameras able to return distance measurements instead of colours therefore they are expensive and have low-resolution. For ultrasonic and infrared sensors, there are categorized in single direction range finders. Some of these sensors require robust coverage while their accuracy is based on the surface they encountered.

Microsoft Kinect Sensor has the combination of an RBG camera, an accelerometer, a motor and a multi-array microphone.

Figure Kinect sensor

This Kinect sensor is released on November 2010 developed by PrimeSense Company in collaboration with Microsoft and its primary usage is for gaming purposes. It is used as an accessory for Xbox 360 console which acts as a gaming controller. The users can use their whole body or body parts such as hands or legs movements where the sensor can understand and converts the information to control and enhance gaming experience.

Aims and Objectives

Scope

CHAPTER 2 LITERATURE REVIEW

Introduction

According to Sigal, L., Balan, A. and Black, M. Humaneva (2012), due to different apprehending method and the model of each human body, the performance in 2D, 2.5D and 3D camp will have differ when the pose is being projected. Therefore, it is essential to note the track. According to Sigal, L., Balan, A. and Black, M. Humaneva (2010), 2D camp is a plan image. This plan image is form from body parts by referring to its model. 2.5D camp is a depth image. This depth image retains the depth contingency model instead of the plan image model. As for the 3D camp, this camp is an image generated in three dimensional form. The 3D image generated shows that as if the human body is form by spherical, rounded and cylindrical parts.

According to Shotton et al. (2011), the Kinect sensor has these characteristics which include:

RGB Camera for facial detection.

IR Emitter and Receiver forms depth sensor.

Software and API owner for example OpenNI.

Skeleton tracking and construction using articulation and joints of human body.

The Kinect sensor able to detects and capture the image and depth of its surrounding environment but it has its own limitation. The information emitted is capture in grid of pixels form and articulations points is calculated by interpolations of intercepted points on grid.

According to Leyvard T, Biometric sign-in tracking and session tracking are use for Kinect sensor identification. He expects to see lots of identification and tracking techniques but with the worries of the accuracy of the Kinect sensor.

Jamie Shotton [ ] uses the depth image to calculate the depth or ‘z’ parameter. The Kinect sensor only provide information on depth but not light or colour information on the depth image. This allows labeling body parts according to body positions.

Christan Plagemann and Varun Ganpathi [11] found out a solution to identify and localize body parts into three dimensional (3D) spaces even though there are depth image together. This method has the advantage where the output can be used directly to conjecture human gesture.

J. Gall [7] proposed a method to capture performance of a human or animal from multi-view video sequence. To find the current pose frame, optimization of skeleton pose is used. Skeleton tracking of the whole body is through tracking the human bone joints. To track a human skeleton, the camera is synchronizes and calibrated according to the human body. Kinect device used in our system is powered by both hardware and software. It does two things: generate a three-dimensional (moving) image of the objects in its field of view and recognize humans among those objects [16].

Chanjira Sinthanayothin, Nonlapas Wongwaen, Wisarut Bholsithi (2012), these are the software that able to install and communicate the Kinect sensor with the computer.

Table 2 - A Comparison Table for Different NUI libraries along with Pros and Cons

OpenNI (Open Natural Interaction) [4] and NITE Primesense [5] is combine to develop codes although OpenNI can be used individually. The main purpose of NITE Primesense Middleware is an image processing, which allows for both hand-point tracking and skeleton tracking. [6]

Flowchart 2 - Flowchart of skeleton tracking and displayed in three dimensions

Kinect sensor has its own limitation on hardware and software. For hardware limitations, according to OpenKinect [26], when the Kinect sensor is exposed to too much sunlight, the depth recognition ability will drop. IR camera is unable to detect the grid projected by Kinect sensor due to the light. The Kinect sensor visual also will be infected by reflective and transparent surfaces because the informative light reflected back will be higher or lower than the expected light. According to Øystein Skotheim [43], objects that are too small and refined will be simplified, smoothed or remain undetected by the Kinect sensor.

As for software limitations, the Kinect sensor is assume to be in a stationary position. Therefore, when the Kinect sensor is on a movable platform, the Kinect sensor may detect more than one user because it may assume that the stationary objects are actually moving and only humans move. Another limitation is the Kinect sensor takes time to calibrate and adjust the virtual skeleton tracking to its user. It will become harder for the Kinect sensor to calibrate when the user is moving.

Kinect Theory

PrimeSense’s 3D imaging system in the Kinect sensor is using a technique called structured-light 3D scanning. Structured-light 3D scanning project many strips of light at the same time and the scanning are based on projection of a narrow strip of light onto a 3D object. When the strips of light reach the object, there will be deformation on the stripes based on the object’s size and shape. From the camera point of view, the stripes projected are different from the original source and this difference is used to measure the distance from each point to the camera and slowly reconstruct the object in 3D volume or 3D image. The camera will provide a large number of samples concurrently as the scanning begins. This method is shown in Figure 2-2-1.

Figure 2-2-1 Triangulation principles for structured-light 3D scanning

The Kinect sensor system uses the similar theory but in some way a different projection and scanning technique. Kinect sensor sensing system is known as IR coding image. The difference is in the projecting where structured-light 3D scanning uses stripes of visible light while IR coding image sends out a pattern of infrared light beams which bounces on the objects and is captured by the standard CMOS image sensor. The image will then be formfitting through an IR-pass filter. Instead of using line patterns, Kinect sensor uses relative positions dot patterns as shown in Figure 2-2-2.

Figure 2-2-2 IR dot pattern emitted by the Kinect sensor.

The combination of both IR emitter and IR receiver is also known as depth sensor. The depth displacements are calculated by using the relative positions of the dots projected by the Kinect sensor in the pattern at each pixel position and return in x,y, and z coordinates of the object.

Figure 2-2-3

The actual depth values are an estimate of the distances from the object to the camera-laser plane rather than the actual distance from the object to the sensor as shown in Figure 2-2-3.

OpenNI Concepts

OpenNI main functional constituent is the production node where node characterizes the sensor’s functionality and high-level abstraction. Sensor functionality includes image or object depth detection, infrared and audio while high-level abstraction includes user positioning, skeleton tracking and hand tracking.

Figure Subclasses of the ProductionNode Class.

Except for ProductionNode Class, there are also five (5) other different classes which are Device class, Codec class, Recorder class, Player class and Generator class. Each class has their own functionality. For example, Device class is for device configuration and Codec class, Recorder class and Player class are for data recording purposes while Generator class epitomizes sensor and middleware features as shown in Figure.

Figure Subclasses of the Generator Class.

From the figure above, table is a brief description of sensor-related Generator subclasses and table is brief description of middleware-related classes.

Generator

Function

Audio Generator

For producing an audio stream.

Depth Generator

For creating a stream of depth maps.

Image Generator

For creating colored image maps.

IR Generator

For creating infrared image maps.

Table

Generator

Function

Gestures Generator

For recognizing hand gestures, such as waving and swiping.

Hands Generator

For hand detection and tracking.

Scene Analyzer

For separating the foreground from the background in a scene, labeling the figures, and detecting the floor. The main output is a stream of labeled depth maps.

User Generator

Generates a representation of a (full or partial) body in the scene.

Table

Kinect Based Humanoid for Rescue Operations in Disaster Hit Areas

Kinect sensor enable human to control humanoid robot using gesture such as human skeleton tracking. In a situation like this, a flexible and easy to control robot is very useful to do on-site rescuer for the ground station far away. The humanoid robot has real-time visual coverage of its environment and the ability to detect abnormal temperatures while searching for survivor.

At the ground station, the Kinect sensor is in position with its user. The Kinect sensor will send gesture information which contains the joint position of the user to the humanoid robot that the user is controlling. The humanoid robot receives the information from the ground station will mimic the movements of the user and perform difficult tasks of rescuing and this will makes the rescue mission more efficient.

The software that used for skeleton tracking simulation is Microsoft Kinect SDK and Microsoft Visual Studio 2010 to do the processing. The humanoid robot limbs are connected with servo motors that have enough torque and able to turn to the angle desired. The total angle that the humanoid robot servos have to turn depends on the user. The joint position information is used to calculate the width of PWM signals that is need from the microcontroller. The program is run at a frame-rate equal to that of the Kinect camera and the angle between these vectors, obtained using dot-product, and gives the angles that the joints of the humanoid should move. [10]

Referances

[4] PrimeSense Ltd., Willow Garage, Side-Kick Ltd., ASUS Inc., AppSide Ltd. OpenNI

TM:Introducing OpenNI, http://www.openni.org/, 2010.

[5] PrimeSense Ltd. NITE Primesense Middleware, http://www.primesense.com/en/nite, 2011.

[6] http://www.cephsmilev2.com/chanjira/Papers/2012/eng/DetectingMovementKinect3DVRIJACT.pdf

[16] How Motion detection Works in Xbox Kinect http://gizmodo.com/5681078/how-motiondetection-works-in-xbox-kinect

[10] http://gimt.edu.in/clientFiles/FILE_REPO/2012/NOV/23/1353648628144/116.pdf

Chapter 3 Research Methodology

Kinect Sensor

The Kinect requires the following computer hardware to function correctly. These are the basic requirements:

A computer with at least one, mostly free, USB 2.0 hub.

• The Kinect sensor takes about seventy percent (70%) of a single hub but not port to transmit its data and information.

• Most systems can achieve this easily, but some palmtops and laptops cannot.

A graphics card capable of handling OpenGL. Most modern computers that have at least an onboard graphics processor can accomplish this.

A machine that can handle 20 MB/second of data depending on how many Kinect Sensors are used. The total data are multiplied by the number of

Kinects you’re using. Modern computers should be able to handle this easily, but some netbooks will have trouble.

A Kinect sensor power supply if your Kinect came with your Xbox 360 console rather than standalone.

The Angular Field-of-View should be 57ºhorizontal and 43°vertical.

Microsoft Kinect sensor usually comes with Microsoft Xbox 360 console but it is possible to get the Kinect sensor individually but make sure that a suitable AC adapter with standard USB connection to connect the sensor and the computer.

Figure 3-1 Kinect power adaptor

Figure 3-2 Kinect external component identification— Output: A) IR (infrared) structured-light laser projector, B) LED indicator, and K) motor to control tilt-in base. Input: F-I) Four microphones, C-D) two cameras (RGB and IR), and E) one accelerometer

Microsoft Kinect Sensor has a depth sensing system which consists of an IR emitter and IR receiver. The infrared CMOS (Complementary Metal–Oxide Semiconductor) sensor is an integrated circuit that contains an array of photo-detectors that act as an infrared image sensor. This device is also referred to as IR camera, IR sensor, depth image CMOS, or CMOS sensor, depending on the source. IR camera also uses a VGA resolution of 640 x 480 pixels with 11-bit depth, providing 2,048 levels of sensitivity and these IR cameras operate at 30Hz. The IR emitter will project a series of light the will be read and recorded by the IR receiver. The data that is being transmitted back are in dots pattern form.

The RGB camera inside the Kinect sensor has its own features. The RGB camera is an 8-bit VGA resolution with 640 x 480 pixels camera. The features include automatic white balancing, black reference, flicker avoidance, colour saturation, and defect correction.

The motor inside Kinect sensor allow it to till its head up and down while the accelerometer determines the position of the head. The motor is a form of small motor driving gears that pitch the tilt of the camera 30 degrees up or down. An accelerometer is a device that measures acceleration. The accelerometer tells the system which way is down by measuring the acceleration due to gravity. This allows the system to set its head at exactly level and to calibrate to a value so the head can be moved at specific angles.

There are four (4) microphones on the Kinect sensor. The microphones are not just for stereo, it’s actually acts as quadraphonic sound system. By combining these four audio inputs, the background noises can be filter out and the relative position of anyone speaking within a room detected. If the Kinect sensor quadraphonic microphones combines with advanced digital signal processing software, these four microphones can be used to do extraordinary things. Microsoft’s official Kinect SDK (Software Development Kit) is the first to reveal how to access the microphones, although other drivers are expected to provide access to this hardware in the future.

Software Integration

OpenNI (Open Natural Interaction) is an open source multi-language, cross-platform framework that defines an API for writing applications utilizing Natural Interaction [3.2].

The main purpose is to make a standard API that enables communication with visual and audio sensors and visual and audio perception middleware.

Figure Abstract Layered View

This OpenNI framework has its own middleware components that do all the image processing.

The middleware components include full body analysis middleware, hand point analysis middleware, gesture detection middleware and scene analyser middleware. All these software are used process Kinect sensor input data.

Full body analysis middleware is used to generate body data analysis that describes joints orientations.

Hand points analysis middleware generates hand point locations.

Gesture detection middleware identifies gesture movements.

Scene analyse scene information for example floor plane coordinates.

Processing

Processing software is the most direct way of interfacing between Kinect sensor and Arduino IDE software. It is a Java-based, open source programming language and IDE. Processing IDE much resembles Arduino’s. Processing is able to talk to Kinect devices through several libraries available from its website. Processing is also capable of talking to Arduino using serial communication.

Processing is a place for its users to create and develop image processing works such as images, animations, and interactions. Processing is an open source programming language and has a suitable environment for all the image processing. Initially, Processing was developed to oblige as a software sketchbook and to teach the first principles of computer programming within a visual context. Processing also has advanced into a tool for engendering finished professional work.

Processing is based in Java, one of the most widespread programming languages available today. Java is an object-oriented, multi-platform programming language. The code you write in Java is compiled into something called bytecode that is then executed by the Java Virtual Machine living in your computer. This allows the programmer (you) to write software without having to worry about the operating system (OS) it will be run on. As you can guess, this is a huge advantage when you are trying to write software to be used in different machines.

Processing programs are called sketches because Processing was first designed as a tool for designers to quickly code ideas that would later be developed in other programming languages. As the project developed, though, Processing expanded its capabilities, and it is now used for many complex projects as well as a sketching tool.

Main Controller

The Arduino board is built around an 8-bit Atmel AVR microcontroller. Depending on the board, you will find different chips including ATmega8, ATmega168, ATmega328, ATmega1280, and ATmega2560. The Arduino board exposes most of the microcontroller’s input and output pins so they can be used as inputs and outputs for other circuits built around the Arduino.

The Arduino Uno has 14 digital input/output pins and six analog input pins, as shown in Figure 1-6. Arduino pins can be set up in two modes: input and output.

Figure Arduino digital pins (top) and analog pins (bottom)

Microcontroller

ATmega328

Operating Voltage

5V

Input Voltage (recommended)

7-12V

Input Voltage (limits)

6-20V

Digital I/O Pins

14 where 6 provide are PWM output

Analog Input Pins

6

DC Current per I/O Pin

40 mA

DC Current for 3.3V Pin

50 mA

Flash Memory

32 KB (ATmega328) of which 0.5 KB used by bootloader

SRAM

2 KB (ATmega328)

EEPROM

1 KB (ATmega328)

Clock Speed

16 MHz

Table

Arduino has 14 digital pins, numbered from 0 to 13. The digital pins can be configured as INPUT or OUTPUT, using the pinMode() function. On both modes, the digital pins can only send or receive digital signals, which consist of two different states: ON (HIGH or 5V) and OFF (LOW, or 0V). Pins set as OUTPUT can provide current to external devices or circuits on demand. Pins set as INPUT are ready to read currents from the devices connected to them. Six of these pins can also be used as pulse width modulation pins (PWM).

The Atmega microcontrollers used in the Arduino boards contain a six-channel analog-to-digital converter (ADC). The function of this device is to convert an analog input voltage also known as potential into a digital number proportional to the magnitude of the input voltage in relation to the reference voltage (5V). The ATmega converter on Arduino board has 10-bit resolution, which means it will return integers from 0 to 1023 proportionally based on the potential you apply compared to the reference 5V. An input potential of 0V will produce a 0, an input potential of 5V will return 1023, and an input potential of 2.5V will return 512.

These pins can actually be set as input or output pins exactly as your digital pins, by calling them A0, A1, etc. If you need more than six digital input or output pins for a project, you can use your analog pins, reading or writing to them as if they were digital pins.

Servo Motor (http://www.robotiksistem.com/servo_motor_types_properties.html)

http://www.anaheimautomation.com/manuals/forms/servo-motor-guide.php

(http://mechatronics.mech.northwestern.edu/design_ref/actuators/servo_motor_intro.html)

Servo motors can be categories into rotary servo motor and linear servo motor. There are three types of rotary servo motor which are brushless DC servo motor, brush DC servo motor, and AC servo motor. The price of servo motor keeps decreasing as it is becoming more widely held and popular. Rotary servos are turning in degrees clockwise or anticlockwise while linear servo motor is flattened out and moves horizontally. The motor of linear servo motor is located inside the plus the coils are positioned outside of a mobile u-channel. For rotary servo motors, except for rotary motion, it is being converted into linear motion. To convert it to linear motion, belts, pulleys and screw thread which include ball screw or lead screw are used.

Servo motor is widely used in robotics industry applications attributable to their affordability, dependability, and uncomplicatedness of being control and coding by microprocessors. The specialty of servo motors is by sending coded signals, the servo motor shaft can be relocated to a desire particular angular position. The angular position of the shaft will change as the coded signal changes. To maintain the position of the servo motor shaft, consistent coded signals have to be applied. The size and shape of the servo motor need depends on its applications. The bigger the torque, the more expensive the servo motor is.

The basic formation of a basic servo motor are DC motor, a gear train or gearbox, a potentiometer, an integrated control circuit, and an output shaft bearing for rotation as shown in Figure

Figure

A servo motor has three wires that incorporate three (3) wires. These three wires are located outside of the servo motor shell.

Figure

The red colour wire is positive (+) signal which is power from four (4) to six (6) volts. Brown colour wire is ground (-) and the orange colour wire is for controlling input signals. The servo shaft will be positioned to a specific angular position when the coded signal is being sent.

Inside the servo motor, the gearbox is being moved by a DC motor with a huge reduction ratio. Force executes at the ending shaft on external load will provide a feedback to the potentiometer. The potentiometer receive the signal allows the control circuitry to determine and observe the up-to-date angle of the servo motor shaft position. Control circuitry will then send the equivalent voltage to an operational amplifier and the voltage will then be compare with the input voltage producing an output voltage. The output servo motor shaft can be position from 0 degree (0°) to 180 degrees (180º). The motor will shuts off and the shaft will remain still if the shaft is positioned at the correct angle. If the control circuitry detects the shaft is not at the correct desire angle, it will power up the motor to turn it towards the correct direction and desired angular position.

Proportional control is used the control the speed of the servo motor. Proportional control means the distance of the servo motor need to travel to its desire angle is proportional to the quantity of power applied to the DC motor. Therefore, the larger distance the shaft needed to turn, the faster the motor speed consuming higher power supply. On the other hand, if the shaft only needs to turn a small angle, the motor will acquire a lower power which causes the motor to turn slower.

Servo motor accurate angular positioning is determined by pulse duration. This is known as Pulse Coded Modulation as shown in Figure. The duration of the pulse width will determine how far the motor will turn that allows the servo to determine the angle of the shaft.

Figure The pulse width of the signals determines how much the servo turns. In the above diagram, 1 represents logic 1 and 0 represents logic 0.

[3.2] http://cs.rochester.edu/courses/297/fall2011/kinect/openni-user-guide.pdf

Chapter 4 Installing Drivers

OpenKinect Introduction

OpenKinect is an open source driver created for Kinect sensor applications. This open source driver is reverse engineered by Héctor Martin. OpenKinect is a low-level driver and this supports Kinect sensor’s motor, LED, camera control as well as accelerometer.

Installation OpenCV

Simple OpenNI Installation

The following steps are required and to be follow closely:

Download and install OpenNI at http://www.openni.org/Downloads/OpenNIModules.aspx with the options of OpenNI binaries, unstable and OpenNI Unstable Build for Windows x 86 (32bit) v1.5.4.0 Development Edition.

Download PrimeSense Sensor Kinect Mod 5.1.2.1 for windows from https://github.com/avin2/SensorKinect/tree/unstable/Bin

Download PrimeSense NITE from http://www.openni.org/Downloads/OpenNIModules.aspx with download options of OpenNI Compliant Middleware Binaries, unstable and build for Windows x 86 (32bit) v1.5.2.21 Development Edition.

After all the above is installed into the computer, plug in the Kinect sensor USB and go to the device manager and update device drivers. If there are newer versions, it is advisable to download according to your computer 32 or 64 bits. There is a software bundle called Mingus. This bundle will install automatically and step by step for OpenNI, PrimeSense and PrimeSense NITE. This method is a very easy step to do but there is a disadvantage because The computer or laptop might not be able to use

Processing Installation

Processing is open source free software that can be downloaded and this software is used mainly for image processing. Processing works very well with OpenNI. Before the Processing software recognizes OpenNI codes, OpenNI libraries must be added into Processing libraries where in this case is C:\Users\Sheryn @.@\Documents\Processing\libraries.

Open Processing and then click import library which is categories under ‘Sketch’. Import library and choose add library. A library manager will pop out and from the library choose SimpleOpenNI. This is a simple wrapper for OpenNI (Kinect Library).

Installing Processing

Go to http://www.processing.org and download

the appropriate version for your OS of choice. If you are a Windows user and you aren’t sure if you

should download the Windows or Windows without Java version, go for the Windows version. The other

version will require you to install the Java Development Kit (JDK) separately. Once you have downloaded

the .zip file, uncompress it; you should find a runnable file.

On Windows, Program Files would seem like a good place for a

new program to live.