Aquifi, Inc.

Teradyne, Inc.

David Demirdjian

Abbas Rafii

Carlo Dal Mutto

Aryan Hazeghi

Eric Lenhart Truebenbach

Michele Stoppa

Giuliano Pasqualotto

Britta Silke Hummel

Ahmed Tashrif Kamal

Pietro Salvagnini

A computing system is provided for training one or more machine learning models to perform at least a portion of a robotic task of a physical robotic system by monitoring a model-based control algorithm associated with the physical robotic system perform at least a portion of the robotic task. One or more robotic task predictions may be defined, via the one or more machine learning models, based upon, at least in part, the training of the one or more machine learning models. The one or more robotic task predictions may be provided to the model-based control algorithm associated with the physical robotic system. The robotic task may be performed, via the model-based control algorithm associated with the robotic system, on the physical robotic system based upon, at least in part, the one or more robotic task predictions defined by the one or more machine learning models.

System and method for predicting robotic tasks with deep learning

A method for a computing device to recalibrate a multiple camera system includes collecting calibration data from one or more pictures captured by the multiple camera system of the computing device. The multiple camera system includes two or more cameras that are each physically separated by a distance from one another. The method further includes detecting decalibration of the camera system. The method further includes, when the camera system is decalibrated, generating recalibration parameters based on the calibration data. The method further includes determining whether the recalibration parameters are valid parameters and, when they are, updating the multiple camera system based on the recalibration parameters.

Multiple camera system with auto recalibration

Dynamic projection of at least first and second patterns contributes detectable disparity onto a scene that includes a target object. The scene is imaged with two-dimensional cameras whose acquired imagery includes disparity contributions whose presence enable a three-dimensional reconstruction depth map to be rapidly and accurately generated. In one embodiment coherent light is input to a first DOE within whose near range output is disposed a second DOE, whose far range output projects an image. Electronically varying effective optical distance between the two DOEs varies the pattern projected from the second DOE. A processor system and algorithms enable dynamic intelligent selection of projected patterns to more readily discern target object characteristics: shape, size, velocity. Patterns can implement spatio-temporal depth reconstruction, spatio-temporal depth reconstruction, and even single-camera spatio-temporal light coding reconstruction. Target objects may be scanned or may make gestures that are rapidly detected and recognized by the system and method.

Dynamically reconfigurable optical pattern generator module useable with a system to rapidly reconstruct three-dimensional data

A method for three-dimensional reconstruction of a scene includes: controlling a projection source to emit invisible light toward the scene; while the projection source is emitting light: controlling a first multi-channel image sensor to capture a first image, the first multi-channel image sensor including visible light detecting pixels and invisible light detecting pixels, the first image including a first invisible light channel and a first visible light channel; controlling a second multi-channel image sensor to capture a second image substantially simultaneously as the capture of the first image, the second multi-channel image sensor including visible light detecting pixels and invisible light detecting pixels, the second image including a second invisible light channel and a second visible light channel; performing stereo analysis of the first image and the second image in accordance with the invisible light channels and the visible light channels to generate a depth map.

System and method for 3D reconstruction using multiple multi-channel cameras

Growing community of inventors

Boca Raton, FL, United States of America

David Demirdjian