Google Inc.

Osaro, Inc.

Rishi Sharma

Frank Pape, Iii

Eric Zhang

Marc M Miller

James Michael McCollum

Brian Schneider

Rachel Werner Barton

Ali Akhavan Bitaghsir

Jessie Lynne Newman

David Wurtz

Ronald E Schneider

Ben David Goodrich

Adam Besen

Hwi Cheong

Hendrik Mueller

David Catmull

Alexander DeNeui

Himanshu Vasishth

Alex Kuefler

Christopher Correa

Sulabh Kumra

A computer system trains a neural network to predict, for each pixel in an input image, the position that a robot's end effector would reach if a grasp ('poke') were attempted at that position. Training data consists of images and end effector positions recorded while a robot attempts grasps in a pick-and-place environment. For an automated grasping policy, the approach is self-supervised, as end effector position labels may be recovered through forward kinematics, without human annotation. Although gathering such physical interaction data is expensive, it is necessary for training and routine operation of state of the art manipulation systems. Therefore, the system comes “for free” while collecting data for other tasks (e.g., grasping, pushing, placing). The system achieves significantly lower root mean squared error than traditional structured light sensors and other self-supervised deep learning methods on difficult, industry-scale jumbled bin datasets.

Computer-automated robot grasp depth estimation

Systems and methods are disclosed herein for opening files via local application. A first application hosted by a local device presents a user interface associated with the first application, the user interface permitting a user to interact with documents on a cloud computing service, receives, via the user interface associated with the first application, a request to open one or more of the documents, and communicates the request to open the document to a second application hosted by the local device, wherein the second application is a local application different from the first application, wherein the second application has access to local information about a plurality of document processing applications installed on the local device, and wherein the request, received via the user interface associated with the first application, to open the document results in the identified document processing application downloading and opening locally a file that corresponds to the requested document on a cloud computing service.

Opening local applications from browsers

Systems and methods are disclosed herein for opening files via local applications. A first application on a local device receives a request to open a document specified by a user via a user interface associated with the first application, the document having a document identifier and associated with a first file stored on a server. The first application communicates the request to open the document associated with the first file to a second application on the local device, and receives, from the second application, information identifying one of the plurality of document processing applications that are on the local device and are capable of opening a second file that is stored on the local device and has the same document identifier as the document specified by the user, the second file being a local copy of the first file. The first application causes the identified document processing application to open locally the second file that corresponds to the requested document associated with the first file stored on the server.

Systems and methods are disclosed herein for opening files via local applications. A first application on a local device receives a request to open a document specified by a user via a user interface associated with the first application, the document having a document identifier and associated with a first file stored on a server, the request comprising the document identifier and a user identifier. The first application forwards the request to open the document associated with the first file to a second application on the local device, and receives, from the second application, a list comprising one or more document processing applications that are on the local device and are capable of opening a second file that is stored on the local device and has the same document identifier as the document specified by the user, the second file being a local copy of the first file. The first application selects one of the document processing applications from the list, and causes a selected processing application to open the second file.

Systems and methods are disclosed herein for opening local applications from browsers. A request to open a document associated with a cloud computing system is received from a user. A determination is made that the local device has a copy of the document, and a local application is selected to open the local copy. A message is transmitted to open the local copy with the selected local application.

Aspects of the present disclosure are directed to architectures, methods and systems and structures that facilitate the sharing and synchronization of electronically stored files among and between cloud entities and a number of computers, systems, devices and/or users. One particular exemplary architectural aspect includes the concurrent determination of file system changes within a cloud file system and a client file system, the serial ordering of necessary file system operations in response to the determined file system changes, and the concurrent execution of file system operations such that the cloud file system and the client computer file system are synchronized.

Sharing and synchronizing electronically stored files

Growing community of inventors

San Francisco, CA, United States of America

Rishi Sharma