Systems DEVELOPING AUTONOMOUS STREET-LEGAL VEHICLES: ANALYSIS OF - - PowerPoint PPT Presentation

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited. 7/23/2018

DEVELOPING AUTONOMOUS STREET-LEGAL VEHICLES: ANALYSIS OF INTELLIGENT GROUND VEHICLE COMPETITION (IGVC) SELF-DRIVE/AUTO-NAV CHALLENGE VEHICLE DESIGN AND IMPLEMENTATION

Andrew Kosinski, Mechanical Engineer, US Army TARDEC Kiran Iyengar, Electrical Engineer, US Army TARDEC

Autonomous Ground Systems

7/23/2018 2

IGV IGVC C C Competit

• mpetition

ion Pur Purpose pose

Objective: The objective of the competition is to challenge students to think creatively as a team about the evolving technologies of vehicle electronic controls, sensors, computer science, robotics, and system integration throughout the design, fabrication, and field testing of autonomous intelligent mobile robots. Educational Benefits: This competition has been highly praised by participating faculty advisors as an excellent multi-disciplinary design experience for student teams, and a number of engineering schools give credit in senior design courses for student participation. Real-world Applications: To advance and promote intelligent mobility for civilian and military ground vehicle

• applications. Intelligent mobility will provide the driver aids required for future Automated Highway

Systems (AHS) and Intelligent Transportation Systems (ITS). For military systems, autonomous mobility will enable unmanned combat vehicles to perform high risk operations and multiply the force effectiveness of manned systems. IGVC objectives for military applications focus on goals established in the Department of Defense. IGVC promotes core intelligent mobility competencies in perception, planning, actuation and mechatronics.

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited. 7/23/2018 3

26 Y 26 Year ears s and R and Running unning

500+ Teams 80+ Universities 7 Countries

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

2018 P 2018 Par artici ticipa pating ting Sc Schools hools

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Compet Competiti ition

• n His

Histor tory

1993 - 2012 Autonomous Challenge 1995 Design Competition 1999 – 2000 Road Debris Course 1999 – 2001, 2003 Follower The Leader 2001 – 2012 Navigation Challenge 2006 – 2013 JAUS Challenge 2013 Auto-Nav Challenge 2014 IOP Challenge 2017 Spec 2 Demo 2018 Self-Drive Challenge

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

IGV IGVC 2018 (26th A C 2018 (26th Anniv nniver ersar sary) y) June 1 une 1-4, 2018, Oakland Univ 4, 2018, Oakland Univer ersit sity

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Self Drive Challenge Design Specifications

Entries must conform to the following specifications :

• FMVSS 500 Platform
• Design:

Side by Side 2-person four-wheel ground vehicle

• Type of Vehicle:

Electrical, no gas

• Maximum Length: 115 in (Polaris Gem e2 is 103 in, Renault Twizy is 91 in)
• Maximum Width: 60 in (Polaris Gem e2 is 55.5 in, Renault Twizy is 47 in)
• Maximum Height: 75 in (Polaris Gem e2 is 73 in, Renault Twizy is 57 in)
• Maximum Weight:

1500 lbs

• Maximum Speed: Speed is limited to 5 mph in 2018
• Speed will increase as safety features of Self-Drive course are developed.
• Mechanical E-stop Location: The E-Stop button must be a push to stop, red, one inch dia.
• Wireless E-Stop must be effective for a minimum of 100 feet.
• Vehicle E-stops must be hardware based and not controlled through software.
• Safety Light: must have easily identified brake lights red in color and reverse lights yellow
• A strobe light mounted on roof and activated when the vehicle is under robotic control.
• Teams may build their own drive-by-wire kits or use off the shelf drive-by-wire solutions :
• TORC Robotics,
• Dataspeed,
• AutonomousStuff
• Clearpath Robotics.
• FMVSS-500 Vehicle Example - Polaris GEM e2

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Self Self-Dr Driv ive Challenge e Challenge at t IGV IGVC C 2018 2018

A freelance event and awards given to the 1st-6th place performance based on performing the following actions: Lane Keeping, Lane switch, Merging, Avoiding crossing obstacles (pedestrians (mannequins) and vehicles), self-parking, right/left/intersection detection/logic. Vehicle Details:

• FMVSS-500 Vehicle equipped with automotive systems drive-by-wire
• Roll bar, seat belts and occupant protection doors or door strapping for safety driver
• Fire extinguisher mounted near safety driver
• External kill switches both sides of vehicle

Sensor Details: Automotive ADAS sensors, Navigation sensor

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

2018 Self-Drive Challenge Video

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

2018 L 2018 LTU TU Self Self-Dr Driv ive V e Vehic ehicle De le Development elopment

• Camera, Radar, Ladar, GPS
• Applying deep learning to lane following
• Various runs/conditions within LTU campus

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

> 0

Smooth Control Law

Lyapunov Stability Method Define the positive definite function as a Lyapunov candidate We seek Speed 𝑤 & 𝜕 that produce Steer 𝜀 that yield Distance 𝑠 & Orientation 𝜄 such that their velocities 𝑠 & 𝜄 ensures that the derivative

≤ 0 negative definite

Lyapunov Stability Method states that a system is stable if a Lyapunov function 𝑤 can be found such that 𝑤 > 0 𝑏𝑜𝑒 𝑤 ≤ 0.

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Lyapuno punov Sta Stabilit bility y Sing Single le Polaris

• laris GEM 2

GEM 2

• Implementation of Lyapunov Stability smooth

control Smooth Control algorithm into physical world model

• Modified Matlab script Robotics toolbox
• One vehicle used for demonstration of

method

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Benefits Benefits of

• f Lyapuno

punov Sta Stabilit bility T y Test esting ing

• Use of scenarios to demonstrate use of

intuitive ground vehicle user interfaces

• Vehicle imbedded training to reduce

warfighters cognitive burden

• Reduce specialized vehicle training
• Enhance performance ground systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Lyapuno punov Sta Stabilit bility y Leader Leader-Follo

• llower

er Multiple Polaris GEM 2’s This is a Leader-Follower Polaris Gem 2 to show the implementation of the Lyapunov Smooth Control algorithm in a constrained

• scenario. In this this scenario the functions

include Lane-Keeping, Stopping and taking a right turn. There is a similar method for a left turn scenario.

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Benefits Benefits of

• f Leader

Leader-Follo

• llower

er Lyapuno punov Sta Stabilit bility Implementa y Implementation tion Benefits to Warfighter: Trajectory takes away stress of having to do route planning during a hostile situation Additional Benefits: Practical use is for parking, real-time steering has to be taken into account Additionally solves problem of tight spots by taking this constraint out of the picture

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Auto uto-Na Nav Challenge Challenge

A fully autonomous unmanned ground robotic vehicle must negotiate around an outdoor

• bstacle course under a prescribed time while staying within the 5 mph speed limit, and avoiding the
• bstacles on the track.

Judges will rank the entries that complete the course based on shortest adjusted time taken. In the event that a vehicle does not finish the course, the judges will rank the entry based on longest adjusted distance traveled. Adjusted time and distance are the net scores given by judges after taking penalties, incurred from obstacle collisions, pothole hits, and boundary crossings, into consideration.

Aw Awar ard Mo Mone ney: $ $ 25,000 00

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

2018 A 2018 Auto uto-Na Nav Challe hallenge nge R Res esult ults

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Des Design ign Compet Competiti ition

• n

Although the ability of the vehicles to negotiate the competition courses is the ultimate measure of product quality, the officials are also interested in the design strategy and process that engineering teams follow to produce their vehicles. Design judging will be by a panel of expert judges and will be conducted separate from and without regard to vehicle performance on the test course. Judging will be based on a written report, an oral presentation and examination of the vehicle. Design innovation is a primary objective of this competition. Two forms of innovation will be judged: First will be a technology (hardware or software) that is new to this competition; and Second will be a substantial subsystem or software upgrade to a vehicle previously entered in the competition. In both cases the innovation needs to be documented, as an innovation, clearly in the written report and emphasized in the oral presentation. Either, or both, forms of innovation will be included in the judges’ consideration.

Aw Awar ard d Mo Mone ney: y: $ $ 7, 7,500 00

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

201 2018 D 8 Des esign ign Compet Competiti ition

• n Result

esults

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

IOP IOP Challenge Challenge

The Interoperability Profile (IOP) Challenge verifies that teams are using a standardized message suitable for controlling all types of unmanned systems, and is the SAE-AS4 unmanned systems standard, commonly known as JAUS. Teams that completed the challenge will send a request for identification to the Common Operating Picture (COP) once every 5 seconds. The COP will respond with the appropriate informative message and request identification in return from the team’s JAUS interface. After the identification report from the COP, the team entry will stop repeating the request. This transaction will serve as the discovery between the OCU via an RF data link and the vehicle. The vehicle that travels the farthest on the course, or completes the course in the shortest time wins.

Award rd Mo Money: ey: $ $ 7,50 500

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

2018 I 2018 IOP OP Challenge Challenge Result esults

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

2018 S 2018 Self elf-Dr Driv ive Challenge e Challenge Result esults

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Rookie

• okie-of
• f-the

the-Year ear

The Rookie-of-the-Year Award will be given out to a team from a new school competing for the first time ever or a school that has not participated in the last five

• competitions. To win the Rookie-of-the-Year Award the team must be the best of the

eligible teams competing and perform to the minimum standards of the following

• events. In the Design Competition you must pass Qualification, in the AUTO-NAV

Challenge you must pass the Rookie Barrel. 2018 IGVC winner was Boise State University.

Awa ward rd Mon

• ney

ey: $1, $1,00 000

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

2018 Grand Award Results

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

IGV IGVC Puts C Puts Students Students to to the T the Test est

• Control theory
• Power requirements/distribution (battery selection, etc.)
• Cognition
• Machine vision (visual/stereo cameras, LIDAR, etc.)
• Vehicle electronics
• Mobile platform fundamentals
• Vehicle electronics
• Sensors
• Systems integration
• Vehicle steering
• Fault tolerance/redundancy
• Noise filtering
• PCB design/analysis/selection
• Vehicle engineering analysis
• Design, fabrication, field testing
• Lane-following
• Avoiding obstacles
• Operation without human intervention
• Detection and navigation of various obstacles:

slopes, potholes, center islands

• Vehicle simulation/virtual evaluation, deep learning
• Overcoming environmental issues (grass, mud, rain, sun)
• Global Positioning System/waypoint navigation
• Safety design

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Tec echnical hnical Challenge Challenge #1 #1 – Frame/Suspension/ ame/Suspension/Mas Mast t Selection/D Selection/Design esign

• Center of gravity
• Material selection (ex. 80/20 T-slotted aluminum framing)
• Suspension systems
• Prefabricated frames/suspension systems (electric

wheelchairs)

Credit for photos: Georgia Institute of Technology, Indian Institute of Technology Bombay, Oakland University, Roger Williams University, Indian Institute of Technology Madras, Lawrence Technological University IGVC Design Reports

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Tec echnical hnical Challenge Challenge #2 #2 – Power/ er/Ba Batt tter ery/ y/PCB PCB Anal Analys ysis/ is/Se Selection lection/F /Fabrica brication tion

• Motor selection: torque analysis/free-body diagrams
• Vehicle weight, coefficient of friction, # of motors, wheel diameter, etc.
• Power draw determination, voltage considerations
• Vehicle power selection: primarily batteries (Lithium Ion, Lithium Polymer, Lead Acid)
• Printed circuit board design

Credit for photos: Hosei University, Embry-Riddle Aeronautical University, Indian Institute of Technology Bombay, Michigan Technological University, Oakland University Pinguino IGVC Design Reports

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Tec echnical hnical Challenge Challenge #3 #3 –E-Stop/Saf Stop/Safety ety Cons Consider iderations tions/D /Design/ esign/Implemen Implementa tation tion

• Vehicle mounted e-stop
• Wireless e-stop (cut power to motor controllers, etc.)
• Adding additional failsafes: turn off motors if fail to receive commands from the computer
• r wireless joystick after set number of milliseconds (200ms – Oakland University team)

Credit for photos: Rochester Institute of Technology, Georgia Institute of Technology, Louisiana State University IGVC Design Reports

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Tec ech h Challenge Challenge #4 #4 - Mac Machine V hine Visi ision

• n -

Sensor Sensor Se Selection lection/Pr /Proce

• cess

ssing/I ing/Imple mplemen menta tation tion

• Mono/stereo cameras, LADAR, etc.
• Redundancy
• 3-D sweeping
• Processing/integration of sensor feeds
• World map generation - simultaneous localization

and mapping (SLAM)

• Noise filtering
• Scene segmentation/recognition
• Varying obstacle detection - lines, barrels, flags,

potholes, ramps, etc.

• Processing techniques - self-learning approaches

(Artificial Neural Network (ANN), etc.)

Credit for photos: Hosei University, Stony Brook University, Oakland University IGVC Design Reports

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Tec echnical hnical Challenge Challenge #5 #5 – Vehic ehicle le Simula Simulation/ tion/Real eal-Lif Life e Test esting ing

• Simulation advantages
• Vehicle can be worked on simultaneously
• Virtual vehicle can be evaluated many times faster than real-time
• Simulation disadvantage
• Only as good as input data/simplifying assumptions
• Requires efficient data analysis - extract useful performance data
• Can simulate entire vehicle at once, or focus on individual sensors
• Introduce noise
• Growing virtual toolset for simulation, analysis and optimization of real-life system

performance: neural networks, evolutionary systems, deep learning (likely huge potential in near future)

Credit for photos: Georgia Institute of Technology, Indian Institute of Technology Bombay IGVC Design Reports

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

IGVC Organizing & Technical Team

• Professor KaC Cheok: Oakland University, Co-Chairman & Co-Founder
• Bernard Theisen: US Army TARDEC Robotics, Co-Chairman
• Jerry Lane: Great Lakes Systems & Tech, Co-Chairman & Co-Founder
• Andrew Kosinski: US Army TARDEC Robotics, Operations Director
• Steve Gadzinski: Ford (Ret), Chief Design Judge
• Matt Skalny: US Army TARDEC Robotics, Interoperability Chief Judge
• Jane Tarakhovsky: Hyundai Mobis, Self Drive Chairman
• Markhanna McBurrows: Oakland University(Ret), Administrative

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

IGV IGVC Sponsor C Sponsors

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Autonomous Ground Systems

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.

Visi isit t our W

• ur Websit

bsite

DISTRIBUTION STATEMENT A. Approved for public release; distribution unlimited.