RoboSub 2017-2018 Jordan Lankford, Ryan Harty, Jake Hannafious, - - PowerPoint PPT Presentation

RoboSub 2017-2018

Jordan Lankford, Ryan Harty, Jake Hannafious, Daniel Mansfield, Marta Camacho, Moeez Malik, Billy Phillips, Ross Dobitz, Oren Pierce, Jeremy Naeve, Nick Baron O2P/VIP Members: Kaitie Wood, Jake Harmon, Samual McCallum, Angel Sanchez, Al Alothman

Acknowledgements

• Advising Professor: Dr. Anthony Maciejewski
• Graduate Advisors: Christopher Robbiano, Megan Emmons
• Industry Advisors: Dr. Jacob Sauer and Torie Hadel with Ball Aerospace

Distinctive Welding Inc.

Budget and Finance

What is the RoboSub Project?

• Competition put on by the U.S. Navy and

Robonation at the TRANSDEC facility in San Diego California

• Autonomous Underwater Vehicle (AUV)

must be fully autonomous

• No communication with the AUV is allowed

during competition

• All teams are required to build and design

the vehicle from the ground up

Last Year

• Project officially started last year
• Researched parts
• First chassis design
• Thruster Testing
• Began work on vision and sensor components
• Propulsion & Power, Vision & Sensors, and Mechanical

Subteams

This Year’s Team

• Propulsion & Power, Vision & Sensors,

Mechanical, and Controls Subteams

• 11 seniors and 5 underclassmen
• Underclassmen perform testing and validation

role

• Semester goal was to have the AUV in the water

and able to track a line

Mechanical Subteam

Ryan Harty Jake Hannafious Daniel Mansfield

What We Started With

• Testing previous team’s design

○ Thermal generation and dissipation

■ Internal temperature reached hazardous levels

○ Forward and lateral hydrodynamic drag

■ Top speed was below acceptable levels

○ Mass and buoyancy measurements

■ Too heavy to compete when loaded to neutral buoyancy ■ (>160lbs, must be <125lbs)

○ Internal volume and electronics layout measurements

■ Electronics difficult to access and limited in space

Where We Went from There

• Fixed Points and Focus

○ Reduce buoyancy and mass ○ Hydrodynamics ○ Electronics layout and accessibility ○ Increase thermal dissipation

• Improvements

○ Top speed increased ■ ~0.6m/s to >2m/s ○ Net buoyancy % reduced ■ ~67% to <15% ○ Dry weight reduced ■ ~110lbs to ~45lbs ○ Thermal Dissipation ■ No data yet, subjectively better

Where We’re Going Next

• CNC machining chassis v2
• New arm to adapt to new challenges

○ Golf ball grabbing/dispensing ○ Golf ball storage/transport magazine ○ Packaged into fairings

• Torpedo launcher

○ Accuracy improvements ○ Packaging into fairings

• Ballast system

○ Reduce dry weight ○ Integrated mandatory safety system

• Casting fairings for hydrodynamic improvement

Testing Video

Propulsion Subteam

Jordan Lankford Moeez Malik Marta Camacho

Propulsion System Constraints

As per the Robosub Competition Rules:

• AUV must be able to:

○ Sense and maneuver in the area using on board resources ○ Be battery powered with all batteries sealed

• Open circuit voltage of any battery or battery system in the vehicle cannot

exceed 60V DC

• All vehicles must contain a clearly marked kill switch that will remove power

to the motors and send AUV to surface

2016 - 2017 Summary

• Investigating Systems - What are other teams doing?
• Major Part Selection

○ Motors ○ Electronic Speed Controllers (ESC) ○ Battery

• Testing

○ Motor Load Testing: Heat up and power dissipation under load ○ Underwater Motor Testing ■ Outward thrust of motors and power dissipation

• Design

○ Battery Management System (BMS) ○ Printed Circuit Board (PCB)

Transfer of Knowledge - Continuing Tasks

• Battery Management System

(BMS)

○ Design around the power requirements of the larger motors ○ Decided to change course due to lack

• f need
• Propulsion Controls and Feedback

○ Controls Subteam formation

• Propulsion Drive

○ Motor actuation and chassis mobility CAD File of BMS Circuit Layout

Propulsion System Design

Work that was done this semester

• Part Modifications
• Current Monitoring
• Motor Control

○ Direct Control: PS2 Controller ○ Indirect Control: Arduino + Pi interfacing

Part Modifications

HobbyKing 60A ESC HobbyKing 30A ESC

UAV Brushless Motor: Max Power 430 Watts

T200 Thruster: Fwd Max: 11 lbs Bkwd: 9 lbs 2016/17 Thruster: Max Fwd/Bkwd: about 5 lbs

M200: Max Power 350 Watts

Current Monitoring

• Hall Effect sensors will sense the

amount of current drawn by the motors

• Enable the Arduino to alert the

Raspberry Pi to reduce the thrust to prevent the battery from draining too quickly

ACS712 Hall Effect Sensor

Motor Control

Direct and Indirect

Direct Control: Motor Control Testing with Gaming Controller

• Control all 6 motors
• Movements mapped to a specific

button

• Ability to move forward/backward,

left/right, up/down

• Roll, pitch, and yaw
• Allows us to record current draw for

each motor as well as internal temperature

Play Station 2 Gaming Controller

Indirect Motor Control - Arduino + Raspberry Pi

Arduino Mega- Slave controller that receives motor state and PWM values from a Master Controller so as to:

• Attach/Detach motors from

power

• Supply a PWM signal to each

motor dictating motor speed Master Controller

Motor State, PWM1,..,PWM6 Motors 1-6 PWM Power

Disarm (0) Arm (1) Error (2)

Error Clear Current Values

Moving Forward…..

PCB Layout for Propulsion Systems... Raspberry Pi Arduino Mega Hall Effect Sensor

Spring 2018 Main Goal

Complete onboard motor control and actuation with some sensor data, hall effects and Pi-Arduino communications

Look Ma, no hands!

Sensors Subteam

Oren Pierce Ross Dobitz Billy Phillips

There are five things the AUV needs to know 1. Current location 2. Depth 3. Desired location 4. Orientation 5. Audio signals

a. Next semester

Overview

Sensors Hardware

Vision Processing, Primary Controller Nvidia Jetson TX2

Sensors Hardware

GoPro Hero 4 Black Raspberry PiCam Cameras Raspberry Pi Pressure Transducer Precision A/D Converter Depth Measurement

Inertial Measurement Unit (IMU)

Why use an IMU?

• Keep the AUV’s balance

○ Crucial for correct movement

• Feedback

○ Check to see if the commands to the motor are being executed

• Dead reckoning

Dead Reckoning

Knowing where you are relative to where you started How is this done?

• Take in Accelerometer and Gyroscope data
• Change the acceleration axis into a world view

○ Will not use magnetometers

• From the corrected acceleration, we can get position

○ This is the hard part due to noise

Vision - The Eyes of the AUV

• Software

○ Language - Python ○ Operating System - Ubuntu 16.04 (Linux) ○ Image Processing - OPENCV ○ Code is developed on the same platform the AUV will be using

• The Code Repository

○ All code is updated to the RoboSub GitHub ○ All code is documented with Doxygen ○ Used to ensure the code can be found and worked on by multiple programmers with ease

Building an Object “Profile”

• Object Classification

○ Certain attributes are used to describe an

• bject

○ An object can be a buoy, target, line, etc.

• Shape and Location Detection

○ What constitutes a shape? ○ How do humans know that a circle is a circle? ○ AUV uses Normalized Cross-Correlation

• Color Detection

○ To a computer, color is defined on an RGB scale ○ The average color of the object is found from the frame returned by cross-correlation Circle Environment

What is Next for Vision

• Functionality for Multiple Object Types

○ Once code is fully functional for buoys, modify to include lines, targets, gates, etc.

• Add Recursion

○ Current code processes only one frame ○ Code will be modified to process images taken by the camera at a certain frame rate

• Tracking and Updating Objects

○ As the AUV moves, the average color and location of an object will change ○ Has an object already been identified? ○ How many different objects are in a frame at

• nce

The Role of Computer Vision

• Since the cameras and image

processing units are making decisions, they have a job

• Controls tells Vision what information

we need to find and return to them

• Vision carries out their task and returns

feedback to controls

• Think of Controls as the Captain of a

Ship and Vision as a Forward Observer

Controls

• Control System Constraints

○ How does everything fit together

• Goals
• Current Work

○ How to implement our Sensors ○ How we control the AUV

• Next Semester

Jeremy Naeve Nick Baron

Overview

• Controls wasn’t a part of the previous year project
• Became a subteam this year because of progress made last year in both

sensors and propulsion

• Necessity as controls will tie everything together in the sub
• Lots of coding necessary
• Have to have a functioning knowledge of both other subteams

Goals

• Collaborate with sensor to develop an interface for communicating

information between AUV systems

• Design autonomous control hierarchy to streamline information transfer

between sub team systems

• Work with propulsion to reliably control the motors
• Begin development of basic automatic control systems.

Control System Constraints

• We had to create the structure for the team
• There was a need to define all the intercommunications of the AUVs

systems

• Where does controls go?

○ We are working on raspberry pi for now ○ There is a transition to a main processing unit ○ The control board (Jetson TX2) will be the “brain” of the system

Current Work

• Design a robust control system to help meet the requirements of

competition

○ Vision ■ Cameras ○ Vehicle Control ■ IMU ■ Pressure Transducers ■ Stabilization

Implementation of Sensors

• Overall hierarchy of sensors systems

○ IMU - depending on accuracy after conducting underwater testing ○ Cameras ■ Identification of competition targets ■ Used to “zero” our current direction ○ Pressure Transducer ■ Should be able to accurately measure depth ■ Should be able to help account for drift of the sub

How we control the AUV

• Through serial connections we will connect sensors and propulsion
• Using feedback loops and input analysis

○ Know where we are in the process ○ Know what tasks we still need to accomplish ○ Know when the tasks are completed and to return “home”

• Complications so far

○ Identification of multiple targets ○ Decision making algorithms

Next Semester

• Integrate sensors systems with autonomous control
• Generate reliable method of autonomous control
• Get the vehicle in the water running on its own

○ Pool testing ○ Set up a course in Horsetooth Reservoir and dive with the sub

• Test early - test often

○ As the subteams complete their tasks, we will need to test the integration of these systems ○ Develop good testbench setup

Conclusion

• Continuation project with the end goal of

competing in the RoboSub competition

• Mechanical

○ New AUV ○ Continued testing ○ Development of competition implements

• Propulsion

○ New thrusters ○ More efficient system ○ Hall effects ○ Continue towards complete autonomy

• Sensors

○ Decisions made on hardware ○ IMU ○ Vision

• Controls

○ Subteam integration ○ Robust control system ○ Just keep coding

Questions?