Three-dimensional computer vision system for robotic devices

A computing apparatus comprising:

• one or more computer-readable storage media;
• a processing system operatively coupled to the one or more computer-readable storage media;
• and
• program instructions, stored on the one or more computer-readable storage media, for training an artificial neural network, wherein the program instructions, when read and executed by the processing system, direct the processing system to:
• obtain ground truth data for a scene, wherein the ground truth data includes depth information for the scene;
• obtain at least two images of the scene, wherein each of the at least two images comprises a different view of the scene;
• and
• provide the ground truth data for the scene and the at least two images of the scene to the artificial neural network, wherein the artificial neural network is configured to generate a depth map from the at least two images based on the ground truth data.

A method of operating a computer vision system, the method comprising:

• collecting a first two-dimensional image of a region with a first camera;
• collecting a second two-dimensional image of the region with a second camera, wherein the second camera is at a different location from the first camera;
• and
• providing the first two-dimensional image and the second two-dimensional image to a neural network, wherein the neural network is configured to generate a depth map of the region based on the first two-dimensional image and the second two-dimensional image.