Goose Chaperone Project Team name : sddec19-17 Client/Advisor : Dr. - - PowerPoint PPT Presentation

Goose Chaperone Project

Team name: sddec19-17 Client/Advisor: Dr. Randall Geiger Presentation by: Johnson Phan Weston Berg Alec Morris Zhihao Cao Woodrow Scott

Website URL: https://sddec19-17.sd.ece.iastate.edu/

Project Plan

Problem Statement:

Goose Chaperone Project: sddec19-17

The victims: golfers, pilots, homeowners The pest: Branta Canadensis - Canada geese The problem:

• Nesting on private property
• Attacking victims who can’t retaliate
• Damaging/Destroying equipment resulting in millions of dollars wasted

Client’s Request:

• Autonomous robot in outside environment
• Employ non-lethal tactics
• Protect property from geese encroachment

Conceptual Sketch:

Goose Chaperone Project: sddec19-17

Functional Requirements:

Goose Chaperone Project: sddec19-17

Weight/ Stability Energy consumption Movement Motorized Wheels Motor Actuator Type

Functional Requirements:

Goose Chaperone Project: sddec19-17

Logic Board GPS Ultrasonic Infrared Camera

Non-functional Requirements:

Goose Chaperone Project: sddec19-17

• Internal logic platform-independent where possible and scalable

○ Platform-independence for future proofing ○ Scalability between different sized prototypes

• Robot can continuously operate for at least 60 minutes

○ Proof of concept not market ready design

• GPS Module frequency should be 5hz or greater

○ Allows for updates after 2 feet of movement.

• Keep cost within $400 budget

○ Set for this project

• Abide by relevant laws

Technical/Other Constraints/Considerations:

Goose Chaperone Project: sddec19-17

• System resources

○ Processing speed, RAM/flash memory, # ports

• Image resolution

○ Too low = not clear, too high = more resource intensive

• Weather conditions

○ Robot’s application is outdoors

• Laws

○ Migratory Bird Treaty Act of 1918

Potential Risks & Mitigation:

Goose Chaperone Project: sddec19-17

• Deterrence method harming people and/or environment

○ Mitigation: Method is low risk, targeting mechanic accurate

• Robot travelling outside target zone

○ Mitigation: Only move if GPS is online

• Software bug causing undefined behavior

○ Mitigation: Proper fail safes, comprehensive software testing

• Hardware failure causing undefined behavior

○ Mitigation: Proper fail safes

• Aggressive geese/animals attack and damage robot

○ Mitigation: Chassis protects internals, effective deterrence methods

Market survey:

Goose Chaperone Project: sddec19-17

• Similar Products

○ ‘Rover-like’, mobile platforms ○ Partially and fully autonomous ○ Deterrence hardware

• What sets our design apart

○ Accurate targeting instead of blanket usage ○ Custom pathing instead of random roaming

Resource/Cost Estimate:

Goose Chaperone Project: sddec19-17

Item name Number Cost Total 2” PVC Tee 5 $1.39 $6.95 2” PVC 90 Elbow 9 $1.09 $9.81 2” PVC Cross 2 $3.29 $6.58 2” x 5’ Long Pipe 1 $9.99 $9.99 Geared DC Motor 2 $24.95 $49.90 Small-Scale Prototype 1 $17.95 $17.95 Item name Number Cost Beaglebone Black Rev C 1 $55.00 Logitech C270 1 $39.99 Ultrasonic Sensor 28015 1 $29.99 Adafruit Ult. GPS 1 $37.44 Total for project is: $263.6 Budget: $400

Project Milestones & Schedule:

Goose Chaperone Project: sddec19-17

System Design

Functional Decomposition:

Goose Chaperone Project: sddec19-17

Detailed Design:

Goose Chaperone Project: sddec19-17

Side View Top View Front View Rotational Structure Smooth Material Storage

Technology Platform(s) used:

Goose Chaperone Project: sddec19-17

Controller:

• Beaglebone Black

○ Capable prototyping board with GPIO, USB ○ 512MB DDR3, 4GB eMMC Flash, 2 PRU 32Bit Microcontrollers ○ Android/Debian compatible Image Recognition:

• Tensorflow

○ Large community support with large availability of pre-trained models ○ High Speed recognition

• OpenCV

○ Supplements Tensorflow to allow location of detected target GPS:

• Adafruit Module 10HZ
• GPSd service

Operating System:

• Debian

Test Plan: Components Test

Goose Chaperone Project: sddec19-17

• Robot Motors testing
• Analog Sensors testing
• Image Processing testing
• GPS testing
• Scare Technique testing
• Goose Identification testing

Test Plan: Prototype test

Goose Chaperone Project: sddec19-17

• Compatibility testing
• Microcontrollers testing
• Data flow testing
• Navigation testing
• Robot movement testing
• Safety testing
• Behavior testing

Prototype Implementations:

Goose Chaperone Project: sddec19-17

• Small Scale

○ ~5 square inches

• Focus on non-structural components

○ GPS ○ Image Recognition ○ Motor Control

• Software developed will:

○ Work with full scale system after calibration ○ Drive larger motors in similar fashion to prototype ○ Will not focus on deterrent apparatus

• Testing Environment

○ Printed images of geese, non-geese ○ Obstacles ○ For image recognition and GPS tests

Conclusion

Current project status:

Goose Chaperone Project: sddec19-17

• The prototype design has been drawn

○ 3 designs ○ 1 selected as prototype

• The materials have been selected

○ PVC materials and cost found ○ Technology and devices found

• The cost has been analyzed and estimated

○ Total cost within budget

Task responsibility:

Goose Chaperone Project: sddec19-17

Westion Berg: Chassis research / Chassis construction / Software for controlling movement Johnson Phan: Drawing prototype design / PVC structure cost and material / Beaglebone sample codes Zhihao Cao: Sensor research / Distance sensor / Image camera Woodrow Scott: Image recognition, software environment and integration Alec Morris: Assisting in GPS integration as well as algorithm analysis/development.

Plan for next semester:

Goose Chaperone Project: sddec19-17

• Over the summer: proof of concept tests on specific modules.
• Within the first few weeks: have a small scale prototype.
• By midpoint: complete construction of larger scale robot.
• By mid-November: deploy path learning algorithms and GPS

usage

• By end: fine-tuning and ensure all requirements are met.

Additional Information

Tensorflow/OpenCV

• Tensorflow

○ Pretrained neural network models ○ Existing models may be quickly retrained ○ Possible to train other behavors

• OpenCv

○ Determine coordinates of match within frame

• Both have public free licenses for commercial use
• Can be used for quick processing in bursts to save battery

Motor Justification

• S : Max slope = 15°
• W : Weight=40lb=18kg
• V : Speed=1m/s
• D : Wheel radius = 10inch = .254m
• Rv = 60V/(𝞀D)=(60*1m/s)/(𝝆.254m)=75rpm
• Ft = gSW = (9.81m/s^2)(.15)(18)=26.487W
• P = FtV = (26.487W)(1m/s)/2=13.24W per motor
• T = (½)(D/2)(Ft)=½*.254/2*26.487=1.24ft-lb=17kg/m