MRSD 2013-2014 Team H: Rev Rev Dennis Cahill Samarth Rajan Gowtham Ranganathan Fan Zhang

IVO

A revolution in revolution…

Introduction

Principle of Operation

Problem statement Build a low-cost, fast, maneuverable, singularity drive based robot. Application Every day 1,000 Americans suffer cardiac arrests. 90% will die unless an Automatic External Defibrillator (AED) is applied to restart the victim’s heart within 3 minutes. We propose a dynamic mobile platform capable of delivering urgent medical equipment in office-like environments.

(a)

Travel at high speed

Get to emergency ASAP

Out of Page

Singularity Drive

(b)

Fast Speed Into Page

Out of Page

(c)

(d)

Direction of Singularity Drive

Fast Speed

Resultant Robot Motion

Singularity Drive • 12 m/s top speed • Brushless DC Motor - 3000 RPM/V @11.1V - Arduino Controlled - PWM • Gear Box - 5:1 Reduction • 3D printed Interface • Hemisphere - Lacrosse Ball 6cm

Performance Metrics Max: 7m/s

Gimbals • Two Nested Gimbals • 2 degrees of freedom • 3D Printed • Servo Motors - Dynamixel AX-12 - Stall Torque 15.3kg·cm - Force Feedback for Traction control - Detailed Diagnostics – Error Sensing - Dynamixel2USB – ROS interface

Singularity Drive Design

Low Cost

Affordable

Gimbals Design

Estimate remaining time

Global Navigation • Generate graph of AprilTags according floor structure • Find shortest path between start tag and goal tag • Tracking AprilTag when detected and guide turning • Utilizing corridor following when no AprilTags detected

Illustration of Corridor Following and Obstacle Avoidance

Legend

Local Planner

0

Global Planner Obstacles

Kinect RGB

Occupancy Grid

<$1500 Drive

Motor

Battery Check

Corridor Following • Extract the center row of the ASUS Xtion (Kinect like device) depth map. • Generate an occupancy grid using ray tracing • Identify highest scoring local path weighted by robot width and turning radius • Send adjusted motor speed signals according to the extension and angle of current path

Motion control • Close-loop direction control using encoders • Traction control to ensure friction and stability

Mechanical Design

Requirements Purpose

Slow Speed

Legend

Use Case Scenario 1. Matt notices Jeff may have gone into cardiac arrest. 2. The RevRev Emergency App detects Matts 911 call and dispatches the Vivo. 3. Vivo speeds down the corridors, avoiding obstacles, and making sharp turns. 4. Vivo delivers a Philips Heartstart AED that saves Jeff’s life.

Name

Vivo’s Range of Motion

Singularity Drive Operation No motion

Software Design (ROS)

< 1/ Minute

Outer Servo Motor Gearbox

Accurate Localisation Global Path Planning

Ensure reliable navigation Create efficient path

Obstacle Detection

Avoid collisions

Global Path Updating

Robust revising path

~99.9%

Inner Gimbal

<2 Seconds

Interface

Inner Servo Motor

Hemisphere

Outer Gimbal

Conclusions

120 Hz <2 Seconds

Results • • • • • •

Robot that averages 1m/s Top Speed: 5m/s Robust obstacle avoidance and corridor following Reliable Navigation between AprilTags Error in destination within 0.5 m Well designed drive system and chassis

• Speed can be further improved using DC motors. • Too many mechanical and software parameters needed manual tuning • We used cheaper components to keep the robot low-cost, so higher quality motors and sensors would likely improve performance • Encoders should be directly attached to the singularity drive to control motor speeds • Less noisy IR sensors or some other sensors should be added to aid the camera, reducing the latency • Single board computers are still fast enough for use in robots