Designing and programming robotics that can manipulate objects is a nontrivial problem, but giving them the ability to “see” their environment and make judgements based on that input is even harder.
In the video above, we talk with Tom Wick, National Project Manager for SICK US, about how robotic vision systems can work and where the technology stands in the market today.
“There’s some programming that’s involved, but we’ve developed a system that’s actually very easy to use,” Wick explained. “We have a vision camera that identifies parts and provides that data to the robot to tell it how to grip it.”
Demonstrated with a UR series collaborative robot from Universal Robots, the programming interface for the camera can be accessed through the teach pendant, by simply plugging in a USB.
“What this allows us to do includes ease-of-use in programming,” Wick continued.
In the video above, we see a vision system mounted separately from a cobot equipped with a gripper end-effector. The vision system is pointed at a small part, which the robot is programmed to manipulate. A demonstrator moves this part to a new location for the robot to locate using the vision camera each time the system resets. What the vision camera sees is displayed on the teach pendant.
“Our system identifies where the part is, using features of the part to track its movement,” Wisk said.
“If we move that part, the teach pendant tells us whether it found the part and tells the robot how far its moved. Once it has identified that, then we teach the positions on exactly how to grab the part to the camera, which is as easy as positioning the robot and calibrating.”
SICK’s vision camera does not use infrared light or a similar technology to see parts, but rather the natural lighting of the room. In the case of the example in the video, the illuminated room and contrast of the black part to the white table provides a clear distinction of the part’s features.
For more information about vision cameras from SICK US, watch the video above and visit the company website.