A109 Team Juggernaut
Team Juggernaut
DARPA Grand Challenge 2005 Technical Paper
Authored by:
DesignJug, LLC
Team Leader: Troy Takach
DesignJug, LLC
Sandy, UT 84092
(801) 947-5729
(801) 907-7870 fax
www.designjug.com
troy_takach@designjug.com
Team Members
Luke Arabasz
Bryan Berrett
Alan Buzbee
Pete Carr
John Christensen
Cameron Clayson
Larry Cluff
Neil Coffey
Phil Conder
Joe Cornwell
Ben Despain
Ben Felsted
Kenton Fife
Cedric Griss
Robert Gunnell
Eun Gyoung Han
Mina Jeong
Bo Jin
Neil Johnson
Jason Jones
Tamas Kerecsen
Taeho Kim
Hyoungsuk (AJ) Kong Brett McRae
Dennis Millard
Robin Pardey
Leonard Peterson
Colleen Shaw
Dave Sorensen
Rangtao Sun
Buzz Takach
Cee Takach
Jean Takach
Joe Van Den Berghe Jim Wheeler
Keith White
Rich Whittaker
Rajiv Yerra
DISCLAIMER: The information contained in this paper does not represent the official policies, either
expressed or implied, of the Defense Advanced Research Projects Agency (DARPA) or the Department
of Defense. DARPA does not guarantee the accuracy or reliability of the information in this paper.
Copyright 2005 DesignJug, LLC
ABSRACT
Team Juggernaut developed a fleet of autonomous ground vehicles for DARPA’s Grand
Challenge 2005. The basis of the development is centered around COTS (Commercial-Off- The
Shelf) technology in order to make the final product affordable, readily available and
understandable by those reasonably skilled in the art. The electronics and software are platform
independent in order to eliminate the need to customize the technology for the various military
ground vehicles subject to retrofit.
Copyright 2005 DesignJug, LLC
2
1. Introduction
DesignJug, LLC believes that autonomous ground vehicles will be a viable and proven
technology within a year. DesignJug research indicates that the capability exists today; it is a
matter of marrying the various technologies currently available with the appropriate software
to create an efficient and cost effective solution. The market exists to support these efforts –
upwards of $10 billion over the next 10 years.
DesignJug has fully committed its resources to the sponsorship of the vehicle - Desert
Juggernaut - as well as the team - Team Juggernaut - for its entry and execution of the
DARPA Grand Challenge 2005. Team Juggernaut consists of full- time and part-time
professionals with over 20 years of experience in ruggedized systems design and
manufacture in a variety of industries. Their backgrounds include: systems design and
development, project management, business management, software engineering, electrical
engineering, mechanical engineering, electronic and mechanical assembly, sales and
marketing, procurement, and strategic planning.
2. Vehicle Description
2.1 Vehicle
The Desert Juggernaut autonomous vehicle is not just one vehicle; it defines a fleet.
These are: V’Ger: a 1997 Yamaha Kodiak all- terrain vehicle whose main purpose is
platform development and proof of concept; Desert Buzzard: a 2005 Polaris Sportsman
MV 7 (Military Vehicle) with its main purpose as a race vehicle and The Jean Marie: a
2004 Chevy Avalanche, both a race vehicle and mobile laboratory. All 3 vehicles have
identical and complete electronics systems and sensors. All 3 vehicles receive software
and hardware revisions as development continues. Redundancy assures that a vehicle will
always be available for continuous development and for race day in the event of some
unforeseen mishap.
For the actual GC’05 National Qualification Event and race, our military ATV was
chosen over the 4-wheel drive truck for two reasons. First, the ATV manages washboard
Copyright 2005 DesignJug, LLC
3
type roads and other “vibration laden terrain” better at higher rates of speed compared to
a 4 wheel drive truck. On this type of road an ATV is not as vulnerable as a truck to loss
of traction. Second, due to the smaller footprint of the ATV, the vehicle is more forgiving
in terms of lateral movement around objects or in the event of slight GPS error
calculation.
V’Ger – a 1997 Yamaha Kodiak. Used for platform development and proof of concept
2.2 Unique Modifications
The military ATV will potentially be modified with a larger gasoline tank to
accommodate the lengthy operational requirements of full day testing and for the GC’05
race day. In addition, a 200 pound steel plate was affixed to the bottom of the chassis to
assist with keeping the vehicle from tipping adversely or rolling over. And finally, a roll
bar system was added to protect the electronics, sensors and structural integrity of the
vehicle in the event of a rollover. The mechanical design was such that the vehicle would
be ‘forced’ to right itself and would not be able to ‘rest’ upside down.
3. Autonomous Operations
Copyright 2005 DesignJug, LLC
4
Our approach is based upon significant experience in delivering and fielding military and
industrial systems for usage in very rugged environments using PC based technologies.
The modularity of the systems and our COTS (commercial-off-the-shelf) orientation
supports our pla tform independence approach, making it ideal for transition into retrofit
production programs for the military. The commercial technologies include sensors,
actuators and industrial computers; all available at affordable costs. Full cycle software
integration has been implemented. In addition, robust mission planning software has been
created.
4. Systems Information
Three (3) 1GHz Pentium computers, running Windows XP, comprise the basis of the
computer system. USB 2.0 was chosen as the connection interface. The software
languages used are based on stable Visual Studio 6 and ActiveX technology.
All data is presented to the various executables through data pipes and shared memory
technology. Seven primary executables are used across two of three functio ning PCs to
process I/O, calculate solutions, and control the vehicle. Those executables are:
Messenger
- Delivers messages from sources to targets
Loader
- Receives all I/O streams into the memory architecture
Vehicle Ctrl - Controls and monitors vehic le operation
DJImageWB - Process all imaging information from sensors
Pathing
- Generates path solutions based upon situational awareness.
Three GPS systems, two cameras, and two compasses determine the orientation of the
vehicle in its environment. All vehicle information is managed through the usage of a
Mission Planning system. Routes from a variety of sources are converted to courses,
assigned to vehicles generating a mission. Data flow is managed throughout the mission
starting with delivery to the assigned vehicle prior to execution and recovery after
execution.
Copyright 2005 DesignJug, LLC
5
The fleet of vehicles can be driven by a human without any modifications, special
instructions or special requirements. All vehicles were driven by humans as needed
throughout the course of development. The Chevy Avalanche truck is used on public
roads on a daily basis.
4.1 System Tests
The Team has implemented a rigorous testing regime. Testing is done across the three
vehicles to validate the robustness of the system. A field mobile laboratory supports
lengthy and extensive road testing. An automated rotary table is used for in- house testing
prior to field testing.
4.1.1 Testing strategy
The GC ’04 course was driven in Spring 2005 to familiarize the team with the
terrain and level of off- road driving difficulty involved. In addition, digital images
of most of the course were recorded for software development purposes. Off-road
and paved-road testing was performed in the western Utah desert, with terrain and
temperatures similar to the GC ’04 course.
4.1.2 Test results and key challenges
Heat was a significant problem when testing in the desert. The electronic systems
did not shut down but computing power was compromised. Due to this, it took
time to troubleshoot the issue back to the electronics and not to a potential
software error. This cha llenge was resolved by installing healthy blowers in all
three vehicles.
5. Summary
Additional information can be obtained by qualified parties upon execution of a non-
disclosure agreement; in consideration of the significant investment by DesignJug, LLC.
Copyright 2005 DesignJug, LLC
6
