The Intelligent Robotics Revolut

DARPA Grand Challenge
The Intelligent Robotics Revolution,
off to a poor start?

The DARPA Grand Challenge autonomous robot race from Barstow California to Primm, Nevada was a dramatic beginning to a very concerted effort by the US military to put robots, side by side, with the Army’s ground forces. Beginning in 2008, the military is scheduled to deploy a full brigade of robot enhanced ground forces which are being developed today in the 25 Billion dollar Future Combat Systems (FCS) program, managed by the Boeing Company.
The purpose of the DARPA Grand Challenge is to entice innovative robot builders, garage shop inventors, university students, and small high tech business owners, to help push this technology from the research labs into the real world. Robot technology is destined to become as seamlessly integrated into our society just like personal computers have done today.
Intelligent Robotic Systems are about to break the same price/performance barrier personal computers did in the mid eighties. In those early days, the typical cost of most computers ranged in the neighborhood of twenty-five to fifty thousand dollars each. With the introduction of the Apple and IBM personal computers and the tremendous increase in the processing power verses the cost of these early machines, computers have become part of the fabric of our everyday life. We are now on the verge of integrating the next technological revolution after the Internet into our everyday lives: Intelligent Robotics.

The coming adoption of intelligent machines (Robotics) will change the global dynamics of our personal and business lives forever more. Manufacturing jobs lost to low cost overseas countries will actually return to the US and other higher standard of living nations such as in Europe and Japan. The costs of producing intelligent robots and automation continue to plummet to their lowest levels ever, while capabilities are simultaneously increasing at a comparable rate. Moore’s Law of computing states that computer power doubles every year, while prices dropped at the same time. Doubling performance every year has successfully predicted the development of personal computers, and soon Moore’s Law will drive robotics technology too.
Intelligent robots will finally become a reality when an Intelligent Operating System is developed, much like Microsoft’s DOS for personal computers in the early eighties. This new robotics operating system will allow these machines to operate intelligently and safely within our environment, requiring far less precision components as used today.
It is clear, by reviewing the entrants to this first ever robot off road race, intelligent robot software has a way to go before intelligent machines become truly commonplace. Never the less, I found the test results and the contestants design concepts very encouraging.

Twenty-five robots were originally entered in the Grand Challenge race. Due to a lack of demonstrable intelligence in the pre-qualification runs held at California’s Ontario Motor Speedway, ten robots were disqualified from the competition. Of the fifteen remaining robots, two successfully traversed seven miles of the course with another machine reaching the five mile mark. The leader was Carnegie Mellon University with their $3.5 million dollar robot entry.
GPS waypoints were released by DARPA to the competitors only two hours before the race. The way points were reported to be 1000 feet apart covering the entire course so collision avoidance using onboard sensors and intelligent behavior was the key to successfully finishing the race. At the last minute, higher resolution GPS systems were dictated by the event sponsor DARPA due to concerns from environmental groups concerned with damage to the wildlife habitat from these autonomous racing robots.
The Cal Tech/JPL robot entry used eight Pentium 3 gigahertz computers, state-of-the-art PC’s by today’s standards, traveled down the 142 mile course just one mile before it encountered a situation the on-board computers couldn’t figure out a clear path for the robot to take. This team spent only about four months developing the current software and they expect their next entry to perform much better in next year’s competition. All of the teams are looking forward to another year of developing their navigation algorithms and critical technology required to solve this very complicated task.
Carnegie Mellon, another university entry, has been working on the autonomous robot navigation problem for at least twenty years longer than anyone else in the world. The team spent five hundred thousand dollars with another three million dollars worth of additional time and equipment donations on their "Sandstorm" robot entry. After the front left wheel was immobilized by the berm of the road, Sandstorm high-centered and plowed smoothly to an even resting place from which it could not move. Sensors, computing and navigation were operating, so Sandstorm was steering and commanding motion to regain the road from its immobile position. This resulted in the wheels just spinning and grinding tire treads right off of their rims. After observing this for a length of time, the chase vehicle hit the emergency stop. Still this first attempt was very impressive. This section of the course was the most troublesome for these intelligent machines and turned out to be the toughest part of the course overall.
DARPA announced the 2004 Grand Challenge a great success and a very important step in enabling robot technology to break out of the labs and plunge directly into our everyday lives. They even suggested this was a history making event, like the Wright Brothers first flight. The Grand Challenge race has definitely opened the public’s eyes to this important advancing new technology. The military services were directed by congress to invest billions of dollars in funding for autonomous robot programs with an eye to deployment later in the decade.
The three primary reasons for failure by the robots in order of importance were: navigation and collision avoidance problems, mechanical breakdowns, and other miscellaneous seemingly minor technical problems. A brake locked up on one vehicle, a drive shaft broke on another, and wires became disconnected due to the pounding and vibration caused by the rough terrain. Simple vehicle malfunctions dogged several entries early in the race which many of us can relate to with our own vehicles.
Cal Tech had a stereo vision system for obstacle detection. The early morning sun caused problems for this team demonstrating how simple things can stop these high tech machines cold. They modified their vision system software to ignore false objects generated by the internal map algorithms when the video cameras looked directly into the sun.
Intelligent robot technology has come forward as a solution for many problems the world faces today. From International trade issues, manufacturing outsourcing to lower cost countries, to the quality of life to the physically handicapped intelligent robots in the very near future will change our lives. Imagine coming home to your personal robot companion which helps around the house. The very difficult navigation and collision avoidance problem will be solved through efforts like the DARPA Grand Challenge robot race and other such events to follow.

DARPA has announced another raced will be held next year with the prize being raised to two million dollars doubling the current prize offered this year. A private group calling themselves the International Robot Racing Federation (IRRF) has already scheduled a similar event later this year and matching the one million dollar prize. This is a spin-off of the DARPA race which was completely overwhelmed with 109 initial contestants which registered to participate in the event.

The ultimate robot race has begun!
BIO-Mr. Golding has been working in the field of Intelligent Machines for about fifteen years. He developed the first Man-Packable robot used in the US Special Forces Program called SOMROV in the mid nineties. The Triune Operating System, or TOS for short, is another of Mr. Golding’s developments which allows a robot or intelligent machine to make intelligent decisions in real-time.

Don Golding
President
Angelus Research Corp.
http://www.angelusresearch.com
dgolding@sbcglobal.net

Copyright 2004
All Rights Reserved