 "Photo: Courtesy of Mahmoud Tarokh")
Computers are so 30 years ago. Robots are the future.
While the past three decades ushered in the rise of the personal computer, in a January 2007 article in Scientific American, Bill Gates suggested that the next 30 years may bring the rise of the personal robot.
Mahmoud Tarokh, a professor in the computer science department at San Diego State, has spent 20 years developing robots and robotic technology. His work has been used by NASA for Mars Rover projects and has been featured in the most prestigious robotics journals.
Until recently, debilitating mathematical problems with robotic technology had stalled the advances in robotics. Researchers finally realized that the problems they encountered couldn't be solved using standard math tools.
"Some of the problems are so complex that if you want to use regular mathematical tools, they might take a million years on a fast computer to get a solution," Tarokh said. "Therefore, those approaches are not useful."
To overcome these problems, scientists started looking at other complex systems, including biological systems, which have many different variables and possible outcomes. One approach is to apply how genetics are coded naturally and make programs that behave in a similar fashion. This style is referred to as genetic algorithms, a set of steps taken to arrive at a solution, similar to how natural evolution does.
In hopes of making more advances, NASA awarded a grant to Tarokh, who had been doing research on genetic algorithms, to see if this technology could be applied to robotic navigation for Mars Rovers. The rocky terrain on Mars was a challenge that could result in one of the small rovers capsizing, effectively ending the mission and any research that could have been done.
Tarokh, who has a Lego model of one such rover on his office desk, developed a solution that allowed the rover's computer to first assess the terrain and then develop a course. He overcame rocky obstacles by writing a mathematical code that broke up the problem into several pieces, like individual chromosomes and genes on a strand of DNA. A computer-generated video showed the culmination of his efforts; a rickety rover dance across the Martian landscape.
Another project at SDSU - initially for use by NASA - is a robot that would follow astronauts, carry tools and transmit images back to Earth. The Robotic Person Follower is being constructed in the Intelligent Machines and Systems lab at SDSU.
Robotic Person Follower utilizes a complex mathematical system in order to function. Like the genetic algorithms used in the Mars Rovers, the Person Follower uses Fuzzy Logic, which is math designed to reach conclusions, like humans do.
You might recognize some everyday applications of Fuzzy Logic: cameras, washing machines and anti-lock brake systems.
However, the Person Follower might be a Fuzzy Logic project that you're not familiar with. It's a one-foot, red-metal box with tires. A computer camera mounted on it, angled up, accounts for the size of an adult human. Tarokh and his team have developed a speaking system that allows it to articulate its thoughts into words.
On a test run around campus and local parks, the robot followed one of Tarokh's graduate students, John Kuo. The run is recorded in a video that shows the robot's ability to work on different terrains and records the feminine-voiced robots thoughts. A non-voice version is available on the professor's Web site.
Set to techno-style music, the video captures Person Follower following Kuo into a GMCS hallway, saying, "My, it's dark in here."
The robot covered terrains such as hiking trails and upon going outside, the Person Follower noticed the change.
"Thank you for taking me out here for a breath of fresh air," it said.
The images showing what the Person Follower actually "sees" is a pixilated circle with a small black spot, representing Kuo's head.
It applies a degree of membership to an object, such as a person, and calculates how much it belongs to a group. Although this is a crude visualization to human eyes, it's effective. In a test, another man walks past Kuo clad in similar clothing and is slightly taller. Even though he walked directly in the path the robot registers a slight change and said, "Hey, are you trying to trick me?" and continues following Kuo.
These life-like robots are already being developed for use in security purposes. One being developed at Tarokh's lab is an ATV - the team stripped down and rebuilt it as a robot with various sensors applying versions of Genetic Algorithms and Fuzzy Logic, and other complex mathematical models.
The goal would be to develop these particular robots so that a team of them could be sent into a building or particular location to secure an area. Tarokh tentatively predicts this goal is not distant and could even occur in a few years.
Gates' prediction may be right on target. It is likely that in the near future, robots won't just be used in space - each family may own one to help with household chores.
Be the first to comment on this article!