4D Vision technology
Artificial intelligence and machine learning power 4D Vision to create new levels of robotic performance and intelligence.
Apera's AI-powered vision technology goes beyond 3D capabilities. Our customers gain speed, intelligence and perception in their workcells, and can do it with simpler hardware than 3D systems.
The scene is captured by 2D cameras and those images are run through our proprietary AI technology. The images are combined into a 3D understanding of the scene, where single objects are identified as pickable.
The robot is given pose estimation and path planning information through the Apera Vue software embedded in its controller. A safe path that avoids collisions is planned. Because the object is trained into an AI neural network using CAD drawings or 3D scans, these steps can happen very quickly.
The object is picked and placed with precise and accurate results.
Training happens in a digital twin environment using synthetic data. Training can happen in parallel to other workcell design activities, cutting down project timelines. The robot, cameras and other equipment are not needed for training.
Our total vision cycle time is lower than any other vision technology—as fast as 3 Hz or 0.3 seconds. A robotic workcell guided by 4D Vision can handle a wider array of objects, including clear, reflective, nested and highly similar parts.
Like humans, Apera AI systems can operate under ambient light without precise fixtures, structured light or lasers. We can hit the industry benchmark of 99% reliability with less hardware and expense involved.
We founded Apera AI with the goal of making robotic automation easier by providing robots with human-like vision intelligence.
Pick objects from simple to complex, including clear and shiny objects, with unprecedented speed and precision.
Task your robot with complex assembly tasks to sub-millimeter precision levels.
4D Vision opens new possibilities in product packaging, including pick and place of multiple item shapes and finishes.
Task robots with sorting multiple objects from mixed bins, including objects that are highly similar.