A robotics and artificial intelligence (AI) developer has received a patent for technology that utilizes robots to train visual and three-dimensional features to computer vision models.
Deep Learning Robotics, a company based in Binyamina-Giv’at Ada, Israel, announced the new patent at the recent AI Week event in Tel Aviv. The company pioneered the use of robots to create accurate training data in 2016, when it first generated and linked visual and spatial features automatically and in real time.
DLR’s systems helped lead to more accurate models, the company said, as well as enabled faster deployments of robotic vision systems — and they could eventually revolutionize how systems generate 3D data. The technology, CEO Carlos Benaim added, could also allow robots to accurately locate and identify objects when using only basic, two-dimensional camera systems.
The company said the patent represents a “testament” to its commitment to pushing the boundaries of robotics and that the company is “excited to continue to explore the full potential of this technology.”
DLR says it is a leader in developing groundbreaking automation solutions, including a robotic controller that allows a robot to learn tasks simply by observing humans performing them.