We have seen amazing displays of artificial intelligence (AI)-inspired robotic systems in everything from board games to autonomous driving, but even the most sophisticated robots are stumped by complex tasks that require the interaction of multiple skills.
Exploring the limits of robotic manipulation, a team from NTU’s School of Mechanical and Aerospace Engineering designed a robot using off-the-shelf hardware, including industrial robotic arms equipped with grippers to pick up objects, force sensors and a 3D camera.
When fed a specific set of algorithms, the robot was capable of assembling an IKEA chair in less than nine minutes, after spending a little more than 11 minutes to independently plan the motion pathways and three seconds to locate the parts.
“For a robot, assembling an IKEA chair with such precision is more complex than it looks,” says team leader Assoc Prof Pham Quang Cuong. “The task has to be broken down into different steps, including identifying the locations of the different chair parts, calculating the force required to grip the parts, and making sure the robotic arms work without colliding with each other.”
The researchers plan to train the robot to assemble a chair from watching a human demonstration, reading the instruction manual, or just by looking at an image of the assembled product.