The Shorthorn - Engineering students ‘play’ with real artificial intelligence

HOME

NEWS

News

Engineering students ‘play’ with real artificial intelligence

Written by Chris Hunt, The Shorthorn staff

Wednesday, 10 March 2010 08:56 PM

Mechanical engineer graduate student Drew Morgan discusses the ground vehicle’s applications and processes Wednesday at the Automation and Robotics Research Institute at the UTA Fort Worth campus. (The Shorthorn: Stephanie Goddard)

Four students play with robots 10 miles away from UTA’s main campus for college credit.

They said this “playing” is essential to the university’s potential upgrade — Tier One status.

The Advanced Controls and Sensors Group lab at UTA’s Automation & Robotics Research Institute in Fort Worth researches cutting-edge technologies that aid humans by developing intelligent, autonomous robots.

Autonomous robots are programmed to adapt to certain situations, allowing them to control themselves without manual control.

“The lab provides graduate and undergraduate students a good understanding of control systems,” said Matt Middleton, computer science and engineering graduate student.

Emanuel Stingu, electrical engineering doctorate student, said the most important advantage of robots is that they come with very little liability compared to a human life.

“They can be used to inspect areas where humans can’t go,” he said.

One of Stingu’s favorite lab toys is the quadrotor. It uses four rotors to hover, autonomously controlling its distance from the ground by listening to an ultrasound sensor on the frame’s underside.

He said innovations like the quadrotor are also used in the film industry to maintain consistency between takes. The lab students achieve this repeatability by developing commands, called algorithms, which allow the robots to operate autonomously.

“One algorithm can make it go the exact same path, every time,” Stingu said.

Drew Morgan, mechanical engineering graduate student, said the next step is to have the quadrotor function completely autonomously, meaning it will have to adapt to changing environments.

“The challenge is to create useful implementations that can be robust enough to function outside a lab,” he said.

The lake behind ARRI is a particularly scary obstacle for these expensive vehicles, he said, and automatic collision avoidance would instill some outdoor confidence.

The Automation and Robotics Research Institute team that is designing the ground vehicle hopes to make it autonomous, or self-regulating. The vehicle was created using parts from an electric wheelchair as the base. (The Shorthorn: Stephanie Goddard)

But research is not limited to the air.

The lab acquired a four-wheel drive scooter, capable of carrying a 400-pound person up to 4 mph. The students’ challenge was to retrofit it with their own components to make it autonomous.

“It’s only about a $12,000 toy,” Morgan said jokingly.

The team then took the shell and seat off, leaving them with a heap of cables, motors and batteries, but no way to control it.

To fix this, the team attached a GPS-enabled circuit board, which allows the vehicle’s paths to be logged for data acquisition.

Wheel and steering sensors give feedback to the circuit board, allowing it to compensate by sending voltage to the necessary wheels.

“It can read feedback from the sensors at 100 times per second,” Morgan said.

The eventual goal is to have a number of vehicles sync together through the use of a common circuit board, he said.

Chris McMurrough, computer science and engineering graduate student, said the ground vehicle will navigate an obstacle course against others at the Intelligent Ground Vehicle Competition this June in Michigan.

Middleton said the ACSG lab’s research is important because it’s not taught in the classroom.

“When you get into the industry, being able to say ‘I’ve actually done this’ is better than saying ‘I’ve learned this,’” he said.