Log in

Google SUV Robotics

Charles Chen, Master Black Belt, Morrill Learning Center, San Jose, CA, USA

Co-Speaker: Mason Chen

Keywords: Lean Six Sigma, DMADOV,  Robotics  

Industry: Education/Training

Level: Basic


During Robotics Science Extra Curriculum Camp, several middle school students conducted a Robotics Science Project by utilizing a Lego EV3 Robot to simulate Google Car on “SUV” fields. Team collected LED Color Sensor data to test whether Robot can follow “SUV” track, and adjust speed at Traffic Lights (Green, Yellow, Red). Students applied Six Sigma and SPSS Statistics to analyze the Robot movement pattern to design & optimize Robot Architecture & EV3 Programming. Adopt Define-Measure-Analyze-Design-Optimize-Verify problem solving roadmap on Robotics Mechanics, Physics, Optics, Programming, Statistics, & Team Building Define Phase Project Objectives 

• Can Lego Robot adjust its speed automatically at traffic lights? 

• Which (S, U, V) field will be more difficult to follow the track? 

• What are the most critical Design factors to impact the performance? 

Project Research Google Cars can adjust its speed at traffic light and are safe like experienced drivers with 75 yrs of driving experience Design Challenges 

• Can we use Lego Robotics to replicate Google Car? 

• Can we outperform Google X Team of 15 superstar engineers? 

• Can we use EV3 free software to simulate Google Chauffeur software? 

• Can we minimize research budget from a Google Car of $150,000 to < $500? 

• Can we use 2 LED Color Sensors to beat Google Velodyne “64-Beam Laser? 

• Can we shorten project cycle from Google > 5 years to < 5 months? 

• Can we design our SUV fields to simulate Google 3D Satellite World Map? 

Alternative Design Conduct Pugh Matrix on Robot Architecture Designs: Color Sensor was selected due to cheaper cost, higher accuracy, and better PID Control over Ultrasonic, Gyro, Infrared sensor Process Flow Chart Design SUV fields (S: many sharp turns; U: 2 90o turns; V: 1 inner tip turn): at green, full speed; at yellow, slows down by 50%; at red light, stops for 5s then full speed SIPOC (Define Project Scope) Begins with Customers (customer-driven) who cares safety. Outputs are what our customers really want: shorter cycle time, better track following, higher reliability & safety. Process is DMADOV Roadmap to optimize performance. 

Inputs are hardware & software Design variables. Supplies are EV3 programmer, and SPSS analyst Team Dynamics Team Building: formed during Robotics Camp, hit “Storming” immediately, standardized DMADOV in Norming & collaborate, respect, mutual trust in Performing Design for Six Sigma Team:

• Follow DMADOV Roadmap 

• Team leader certified IASSC Black Belt, IBM SPSS Statistics 

• Team mentor is ASQ CSSBB, CQE, CRE, CMQ/OE certified 

• Key members are IASSC Yellow Belt Certified & IBM SPSS Statistics Certified Measure Phase Measure System Analysis: improve Gage R&R by fixing the initial Robot placement Baseline Process Capability Analysis 

• Data Collection Plan: set sample size = 7 (resource and budget constraint), and avoid Type II Beta risk of “failing to reject” the Null Hypothesis 

• Scatter Plot: observe two major slow-down hard-turn regions 

• Box Plot: observe several outliers & use Median to present the Central Tendency. S Curve has the worse cycle time than U and V (more Sharp Turns) Set up Performance Metric: Median Cycle Time instead of using Cpk (Team Decision): S < 20s, U < 19s, V < 18.5s Analyze Phase Subjective Root Cause Analysis: conducted C&E and 5 Whys Analysis Why Robot could not follow the track at sharp turns? 

• Robot could not tell where is the track edge? The “White” background color intensity was too close to the Green track color intensity 

• Robot is moving too fast? The Robot mechanics have limited the sharper turning 

• Ambient light intensity is not well controlled: the incoming light may be blocked by Robot Body 

Why Robot cycle time is not consistent? The robot initial placement is critical to the early movement Analytical Root Cause Analysis on the Baseline Raw Data Conduct Histogram Analysis: at sharper turns, Robot slowed down significantly to make such Hard Turns Conduct Contingency Analysis to locate Where the most slow-down regions are: two major Hard Turn areas in S field: (1) 2nd Green Zone, (2) 3rd Yellow Zone) Vital Few Xs Identify top seven critical variables through X-Y Pareto Matrix Analysis. Conduct 2-sample t tests & verify Five Significant Vital Few Xs: Pass Normality Tests (P-value > 0.05) Design Phase Design Strategy: brainstorm a “Pareto” strategy and use Hard Turn as primer Quality Metric to reduce the cycle time Enhance SUV Field to address concern that green color intensity is too close to the white background. Added the “Black Line” (darkest contrast to White) to separate the white background & the Green-Yellow-Red zone Two LED Color Sensors: Color Intensity Mode (Red LED, PID-Control) to follow the black line; Color Number Mode (Green Yellow Blue LED) to follow the traffic light colors Back Wheel Ball Design: front-Wheel Drive and back ball wheel provide easier maneuver allowing more freedom Software Design Principles: define two “Turn” & “Speed” Variables to allow slow down the Robot Movement at Hard Turns determined by “Hard Turn” & “Gentle Turn” Threshold Optimize Phase Design of Experiment Screen Phase: use one factor at a time (OFAT) analysis to filter out any insignificant factor. Reduce input variables from 5 to 3 Optimize Phase: use both Full Factorial and EVOP (Evolutionary Optimization) 

• Identify the optimum setting for both Architecture Hardware & EV3 Programming 

• Improve the predictive model by trimming the higher order terms 

• Model has aligned with previous Root Cause Analysis and Design Strategy Conduct Optimum Process Capability Analysis: all SUV Median meet the performance requirement Verify Phase Design Verification Compare the Cycle Time (Optimum Design vs. Baseline Design), Sample Size= 30 each 

• Conduct 1-sided 2-sample t test at 95% Confidence Level: 2-sample t P-Value < 0.05, > 95% confidence to reject Ho, and draw statistical conclusion that the optimum setting has significantly improved cycle time 

• Confirm Hard Reduction Strategy by Contingency Table Analysis. Verify the most improvement at Hard Turns Finance Effect Analysis and Takeaways 

• SUV Project has successfully improved the Self-Driving Car efficiency and safety concern 

• Reduced research funding from > $150,000 Google Car to < $500 SUV Robot per Unit 

• Shorten the Design and Process Development Cycle time from > 5 years to < 5 months 

• Invaluable learning opportunity for promoting the DMADOV Team-Building Roadmap What will participants learn? 

• Engineering Statistics can be applied to Middle School Science, and Physics Project 

• Six Sigma DMADOV can develop young kids Critical Thinking, Problem Solving, Team Building and Quality Leadership 

• Middle School students have potential to achieve Lean Six Sigma Black Belt level by applying learning on their Academic Knowledge

Participating Organizations at the Lean & Six Sigma  World Conference

Government Agencies

  • Department of Commerce
  • Department of Defense
  • Department of Energy
  • Department of Health & Human Svcs.
  • Department of Homeland Security

  • Department of Justice
  • Department of State
  • Department of the Treasury
  • Department of Transportation
  • Department of Veterans Affairs
  • Environmental Protection Agency
  • NASA
  • Naval Surface Warfare Center
  • Pentagon
  • U.S. Air Force

  • U.S. Army
  • U.S. Marine Corps
  • U.S. Navy
  • U.S. Veterans Affairs
  • United States Army Corps of Engineers


  • AIG
  • Alcoa
  • AT&T
  • Bank of America Corp
  • BASF Corporation
  • Bayer Corporation
  • BMW
  • The Boeing Company
  • Bose Corporation
  • Bristol-Myers Squibb
  • Campbell Soup Company
  • Cardinal Health
  • Caterpillar
  • Chrysler Corporation
  • Chevron
  • Cisco Systems
  • Coca-Cola
  • Comcast
  • Daimler Chrysler
  • Disney
  • Dow Chemical

  • Dr Pepper 
  • Duracell
  • Dupont
  • Eastman Kodak
  • Facebook
  • Google
  • Exxon Mobil
  • Fedex
  • Ford Motor
  • General Electric 
  • General Motors
  • Gillette
  • Goodyear Tire
  • Hewlett Packard
  • Honeywell
  • Humana
  • IBM
  • Kohler
  • Lockheed Martin
  • Macy’s
  • M&M/Mars
  • ManpowerGroup
  • Maytag Appliances
  • Mercedes
  • Merck
  • Mitsubishi
  • Mobil Chemical
  • Motorola
  • NASA
  • Nestle 
  • Northrop Grumman
  • PepsiCo
  • Philip Morris International
  • PNC Financial Services Group
  • Pfizer
  • Pratt & Whitney
  • Procter & Gamble
  • Prudential
  • Raytheon
  • Rolls Royce Allison
  • Target
  • Johnson & Johnson 
  • Schindler Elevator Corporation
  • Schneider Electric
  • Shell
  • Siemens
  • Southwest Airlines
  • Staples
  • Tesla
  • Tiffany & Co.
  • Qualcomm
  • Underwriter Laboratories
  • UnitedHealth Group
  • United Technologies
  • Union Pacific
  • UPS
  • USAA
  • Verizon
  • Walmart
  • Wells Fargo
  • Westinghouse
  • Whirlpool
  • Xerox


Lean & Six Sigma World Conference




Log in

© Copyright 2019 American Quality Institute. All Rights Reserved.