Project Periscope

Design Review

Team:
Aabhas Sharma (asharm36)
Michael Congdon (mcongdo2)
Siddharth Murali (smurali2)

TA:
Ankit Jain

ECE 445 – Senior Design

Fall 2015
University of Illinois at Urbana-Champaign
Table of Contents
1. Introduction………………………………………………………………………………3
a. Statement of Purpose 3
b. Objectives 3
i. Consumer Benefits 3
ii. Product Features 4
2. Design……………………………………………………………………………………5
a. Block Diagram 5
b. Software Flowchart 6
c. Electrical Schematics for Control Unit 7
d. Block Description 10
i. Camera Unit
ii. Computer – Oculus SDK
iii. Oculus Rift unit
iv. Swivel unit
3. Requirements and Verification…………………………………………………………13
a. Modular Functionality 13
b. Tolerance Analysis 16
4. Cost and Schedule………………………………………………………………………18
a. Cost Analysis 18
i. Labor
ii. Parts
iii. Grand Total
b. Schedule 20
5. Ethical Guidelines and Safety Requirements…………………………………………...24
a. Ethical Guidelines 24
b. Safety Requirements 26
6. Appendix……………………………………………………………………………….27

2
1. Introduction:

Title: Project Periscope

● Statement:

Currently there's a lot we can do with VR, but can we actually be in an entirely different

place in a matter of seconds? We’re not quite there yet. The idea behind Project Periscope

is to be able to create a more engaging virtual reality experience - one where we can

actually have eyes in a very different part of the world.

Our plan, in its element, is quite simple – connect a camera to an Oculus Rift. We want to

view a camera’s live feed on the Oculus’ display while being able to physically navigate

the camera based on the Oculus wearer's head movements – give the whole experience a

more realistic feel. What this essentially means is that the camera will follow the wearer’s

movements, allowing a user to not just see a live feed of someplace else, but also be able

to freely navigate through the view with a broader field of vision – all this in real time.

 Objective:

Consumer Benefits

● Be able to view a very different part of the world real time

● Physically navigate through their field of vision - simulating human vision and

free range of movement

● Multiple applications:

3
 ○ Can be used as a remote surveillance system (with mobile capabilities as

well)

○ Can be used for remote imaging - aerial, terrestrial, and marine

○ Exploring the world - see places where you can’t otherwise go to. For

example, visit the Eiffel Tower! A VR substitute for View Master

○ Can bridge telepresence with media events - such as football games and

conventions or conferences like ComicCon - where the user can explore the

event with their own freedom - look wherever they find most interesting

○ Open up future possibilities for nano-imaging, especially for the medical

field, allowing doctors to see inside the human body like never before

● Product Features

● Wireless Streaming

● Wireless data-transfer between GoPro and the Oculus

● First-person remote viewing of virtually any environment

● Remotely controlled camera

● Head motion tracking

4
2. Design:

Block Diagram

Figure 1: top level block diagram showing data and power connections between various modules.

5
Figure 2: Software flow chart detailing process flow between Camera Unit, Oculus Unit and Swivel Unit

6
Figure 3: top level schematic showing basic interconnections of motor controller to drivers and motors

7
 Figure 4: Detailed view of connections to Raspberry Pi

Figure 5: detailed view of motor driver circuit, including peripheral components necessary to regulate voltage and
boost switching capacity

8
Figure 6: buck converter to provide 5 volts for the Raspberry Pi from 12 volt battery

Figure 7: isolation circuit to protect Raspberry Pi from 5v logic level of motor

encoders

9
Block Descriptions

 Camera

 The camera is the “eyes” of our project. It has its own WiFi hotspot, which the laptop

connects to in order to stream the video. The video feed is passed from the camera to an

image processing program, which will stitch the two images together and then pass it to

the Oculus SDK

 Microcontroller 1

 Will be using this to conduct the image processing required to modify and stitch

together the camera feed to align with the Oculus display’s expected format.

 Computer - Oculus SDK

 The Oculus SDK is what controls what you see as well as which way you turn. The video

feed is transmitted from the image processing software into the SDK, and this transmits

the feed into the Oculus display, showing you what the cameras see. The SDK also

receives head movement data from the accelerometer/gyroscope and uses this to

determine the direction in which the motor needs to move.

 Oculus Rift Unit

 This consists of three parts – display, power supply, and motion tracking

 The display gets the video from the SDK and shows it to the user as a 360 degree view.

 The power supply powers the Oculus Rift Unit.

 The motion tracking consists of an accelerometer, a gyroscope, and a magnetometer.

These three combined relay the headset orientation data to the SDK.

10
 Swivel Unit

 Swivel Unit Power Supply

 Inputs: none

 Outputs: 12v and 5v supply rails

 Components:

 lead acid battery

 Selected for substantial capacity and low cost

 Buck converter

 Provides 5 volt supply for Raspberry Pi

 Utilizes LM2675 chip to handle high side switching and feedback

control to ensure steady output

 Orientation Control Unit

 Inputs: orientation data from Oculus SDK, encoder pulses from motor encoders

 Outputs: PWM signals to motors

 Components:

 Raspberry Pi running Simulink modules

 Wirelessly communicates with the computer in order to receive

orientation data from the Oculus SDK.

 Motor Shield and Motors

 Inputs: PWM from control unit

 Outputs: encoder pulses

11
 Components:

 Isolation circuitry

 Logic level converter will protect the 3.3 volt logic of the

Raspberry Pi from the 5 volt logic of the motor encoders.

 3.3 volt Zener diodes will be used as a last line of defense if one of

the LLC’s should fail.

 Motor drivers

 L6203 DMOS full bridge drivers receive pwm pulses from Pi and

drive the motors.

 Brushed gear drive motors with quadrature encoders

 The motors each have 2-pin hall sensor quadrature encoders with a

resolution of 64 pulses per revolution of the motor shaft.

 The gear box has a 70:1 gearing ratio, which gives a total of 4,480

encoder pulses per revolution of the output shaft

12
 3. Requirements and Verifications

● Modular Functionality

Requirement Verification Points

Oculus Rift Unit

● The Oculus position data 1. Compare vectors during application runtime 10

must correspond 1:1 with 2. SDK 3D position vectors (Pitch, Yaw, and Roll)

User’s movements with a and User movements must align to the exact

tolerance of ~10 degrees degree with a tolerance of ± 10 degrees

● Achieve a maximum latency 1. Compare timestamps for corresponding User 10

of 18ms ± 2ms movement degree and 3D Vector degree

2. When they are exactly equal, timestamps should

not differ by more than 18ms with tolerance of ±

2ms

Camera unit

● GoPros need to be at a 1. Measure the distance using the Vernier caliper to 5

separation of 63.5mm with a obtain a measure of 63.5 ± 3.65mm

tolerance of ~ 3.65 mm

13
● Image processor barrel 1. Use test 2D images with available Oculus output 20

function must return images

modified image with exact 2. Compare our processed output with sample

concave depth as in this Oculus image

formula: 0.24r4 +0.22r2 + 1 3. Output should be the same as the sample image,

where r is the distance with minor differences allowed

between any pixel and the

center of a rift lense2

● Tolerance ~2px

● To meet our latency 1. Run program on Raspberry Pi. Make note of 15

standards of 18ms(± 2ms), program start time

the image processing must 2. When program finishes, the timestamp at the end

return output image within of the program must be within 15 ± 5ms of the

15ms ± 5ms program start time

Swivel Arm Unit

● Buck converter maintains 1. Power Raspberry Pi with buck converter 10

efficiency of 90+% while 2. Have Pi power 6 LED’s through GPIO pins to

microcontroller is operating simulate load of driving motor drivers

14
 3. Measure input and output voltage and current

of buck converter, ensure adequate efficiency

rating is met

● Motors are able to produce 1. Attach simulated load calculated to require 1.5 10

1.5 Nm of torque to within Nm of torque to lift/rotate

10% 2. Power motors with battery, make sure they are

able to lift the load

● Motors should be able to 1. Power on system, with either the real camera 20

actuate camera unit with unit or simulated loads attached to each motor

less than 7% overshoot. (At 2. Using step input (to simulate worst case

8+%, the differences in scenario of infinitely fast head movement),

FOV’s may be overcome have motors turn unit 180 degrees

and the video will oscillate) 3. Measure encoder outputs and analyze system

response. Should not exceed 7% overshoot.

Total 100

15
● Tolerance Analysis

A major challenge for us to be able to map a 2D camera image feed to the concave

Oculus display. The Oculus offers the user a very immersive experience by presenting a

virtual image that appears larger than it actually is. Although this is a key positive aspect

that enables the complete VR experience, it could result in major health hazards if not

done right. If the Oculus’ display is screening images that are distorted or have

unexpected skips in motion, the user could become vulnerable to seizures, epilepsy

attacks, motion sickness etc.

Given this situation, it is important for us to ensure we are significantly accurate in our

image processing. We will be running our camera feed through barrel distortion

algorithms2 that will map our two dimensional feed to a concave screen using the

following formula: 0.24r4 +0.22r2 + 1. Basically how this works is that for any pixel on

the Oculus screen that is distance r away from the centre of the concave screen, the

appropriate pixel to be copied over from the 2D camera feed is given by 0.24r4 +0.22r2 +

1. 2

Our tolerance to this has to be significantly small. As highlighted earlier, we can’t have a

poorly distorted image. Therefore, we are only accepting a tight tolerance of ± 2px offset

in radius distance while mapping the 2D image to a barrel image.

The way we will ensure this tolerance is using a fundamental method of verification that

is used in most image comparison software - we will use sample Oculus images as the

16
input for our image processor. We will then compare our output image to the Oculus

version of the sample image. While comparing, we will not just check each 2D pixel with

its corresponding barrel pixel location, we will check all barrel pixels within a 2 pixel

radius of the expected barrel location. Our verification program will run through the

entire output image and compare each pixel with nearest neighbors (2px) on the sample

output, if there is a match, we would consider this phase a success, else we will have to

go back and debug.

17
4. Cost and Schedule:

Part Value Digikey # Price Qua.

Buck Converter (x1)
Capacitors 4.7 uF 493-10470-1-ND $ 0.24 1
10 nF 490-1312-1-ND $ 0.10 1
100 nF 490-1318-1-ND $ 0.10 1
47 uF P5539-ND $ 0.23 1
inductors 56 mH AIUR-06-560K-ND $ 0.88 1
diode Schottky MSS2P3-M3/89AGICT-ND $ 0.37 1
IC LM2675 LM2675MX-5.0/NOPBCT-ND $ 4.03 1
$ 5.95

Motor Driver (x3)

resistors 150 ohm 311-150GRCT-ND $ 0.10 1
10 ohm 311-10ERCT-ND $ 0.10 1
0.5 ohm 311-0.5LWCT-ND $ 0.47 1
capacitors .22 uF 493-1096-ND $ 0.23 1
100 nF 490-1318-1-ND $ 0.10 1
15 nF 490-1643-1-ND $ 0.34 2
22 nF 490-3884-1-ND $ 0.10 1
zener 12 v 568-6354-1-ND $ 0.20 1
IC L6203 497-1421-5-ND $ 9.53 1
$ 11.51
$ 34.53

Logic Level Converter (x12)

resistors 10 kohm 311-10.0KCRCT-ND $ 0.10 2
IC BSS138 BSS138CT-ND $ 0.23 1
zener 3.3 v MMSZ4684-TPMSCT-ND $ 0.15 1
$ 0.58
$ 6.96

Microcontrollers, etc…
controllers Raspberry Pi $ 50.00 2
vision Oculus Rift $ 350.00 1
cameras GoPro Hero3+ Silver $ 300.00 2
motors 70:1 Metal Gearmotor $ 40.00 3
battery sealed lead-acid $ 20.00 1
battery charger UPG D1724 $ 9.47 1
$ 1,199.47

Total cost of parts $ 1,246.91

18
Name Rate Time Cost
Siddharth Murali $ 75.00 200 $ 37,500.00
Aabhas Sharma $ 75.00 200 $ 37,500.00
Michael Congdon $ 75.00 200 $ 37,500.00

Total $ 112,500.00

Section Total
Parts $ 1,246.91
Labor $ 112,500.00
Total $ 113,746.91

19
Schedule

Week Task Responsibility

30-Sep Design circuitry for motor controller Michael

Obtain and set up Oculus Rift and SDK Siddharth

Get raw video feed from camera to computer Aabhas

7-Oct Design armature, meet with machine shop Michael

Design microcontroller algorithm for image Aabhas

processing

Implement Oculus application for viewing raw video Siddharth

feed from a camera

Order motors, give them to machine shop. Finalize Michael

14-Oct PCB design

Program Oculus for transmitting head motion data. Siddharth

Cross reference parts list for ECE shop availability.

Implement microcontroller algorithm for image Aabhas

processing

20
21-Oct Characterize system dynamics, write control logic Michael

for Simulink

Implement protocol to receive head motion data Siddharth

from Oculus and Simulink on a Pi

Optimize image processing algorithm to return Aabhas

output within tolerance timeframe

28-Oct Test/validate motor function and basic control Michael

Test/validate Oculus functions (display, head Siddharth

tracking)

Design microcontroller algorithm for image Aabhas

processing

Program Raspberry Pi for transmitting video to

Oculus

4-Nov Refine control logic Michael

Tolerance testing of system Siddharth

Test/validate video transmission and image Aabhas

processing. Prepare for Mock Demo

21
11-Nov Final assembly of components Michael

Ensure mobility of system Siddharth

Ensure entire system functionality Aabhas

18-Nov Final testing and debugging – Hardware Michael

Final testing and debugging – Oculus + Motor Siddharth

Control

Final testing and debugging – Software (Camera) Aabhas

25-Nov Make sure project is ready for demo Michael

Make sure final presentation meets format Siddharth

requirements

Create presentation support content and talking Aabhas

points for team members

2-Dec Create final paper content for Swivel unit and ensure Michael

uniform formatting

Create final paper content for Oculus unit and ensure Siddharth

uniform formatting

22
 Create final paper content for Camera unit and Aabhas

ensure uniform formatting

9-Dec Make sure all major parts are returned Michael

Make sure nothing is missing from lab kit Siddharth

Complete project closure related tasks such as IPR Aabhas

management and secure code sharing

23
5. Ethical Guidelines and Safety requirements

a. Ethical Guidelines

We will adhere to the following guidelines from the IEEE Code of Ethics as they are relevant

to our project in the capacity listed below.

1. To accept responsibility in making decisions consistent with the safety, health, and

welfare of the public, and to disclose promptly factors that might endanger the

public or the environment;

We will be applying high safety standards to ensure that our product will be easy to use.

We will have warnings that list the potential consequences of using our product for too

long, and safety methods to be followed to ensure that the user has a enjoyable and safe

experience with the product.

2. To be honest and realistic in stating claims or estimates based on available data;

We will cite all sources used in formulating our claims, and we will mention the areas

where we have chosen to make our own claims and give our reasoning behind them.

3. To maintain and improve our technical competence and to undertake technological

tasks for others only if qualified by training or experience, or after full disclosure of

pertinent limitations;

We will have taken all the required courses in order to complete this project and we will

do additional research in the areas that our expertise is lacking.

24
4. To seek, accept, and offer honest criticism of technical work, to acknowledge and

correct errors, and to credit properly the contributions of others;

We are working closely with the TAs and professors in the Design Review and peer

evaluations of our project, and we will work to incorporate their criticism and suggestions

into our final design.

5. To treat fairly all persons and to not engage in acts of discrimination based on race,

religion, gender, disability, age, national origin, sexual orientation, gender identity,

or gender expression;

We are a team and we will distribute work and give honest evaluations of team members

without discriminating against anyone. If there is a problem we will meet with the

TA/professor and work it out.

6. To avoid injuring others, their property, reputation, or employment by false or

malicious action;

We do not wish to cause harm to any person or their property in any way. Our product is

intended to broaden the scope of a particular area and is not intended to harm anyone

intentionally. We will mark all dangers of our product with safety warnings if required.

7. To assist colleagues and co-workers in their professional development and to

support them in following this code of ethics.

We will do our best to support other teams and projects in this class by offering help that

is within our area of expertise. We will participate in peer design reviews in order to aid

this process.

25
Safety requirements

● There are a few common safety hazards connected with the usage of an Oculus

Rift. Before using this system, go through the safety documentation created by

Oculus Rift (Link).

● Pinching Hazard - Our product has motors that will rotate depending on user

movement. If someone gets their finger caught it could cause injury. Individuals

need to be careful while working with the motors and the camera mount as they

may have pinch points. Also keep your hands away from the motor when the

device is in use as that is the most likely to cause injury.

● Our product has many components that cannot be exposed to water. If they are,

there is the chance of them shorting out and destroying the circuit or injuring the

user.

26
Appendix:

1. Angular velocity and acceleration data acquired from this study:

http://scholar.lib.vt.edu/theses/available/etd-08182005-222028/unrestricted/thesis.pdf

2. How barrel distortion works

https://www.youtube.com/watch?v=B7qrgrrHry0&feature=youtu.be&t=11m26s

Barrel Distortion

http://stackoverflow.com/questions/28130618/what-ist-the-correct-oculus-rift-barrel-

distortion-radius-function

27