Abstract: A device comprising a hand-held manipulator and a camera assembly configured to capture a field of view encompassing a workspace centered on a forward-projecting axis and at least one visual feature angularly displaced by at least 90 degrees from the forward-projecting axis. The wide-angle field of view enables robust spatial tracking and mapping even when the workspace is occluded or the manipulator is operating in confined environments. A video processing system processes the video data to generate training information for robotic manipulation tasks, using bimanual Simultaneous Location and Mapping (SLAM) and other computer vision techniques.

Abstract: Examples relate to a system for generating robot training data. A gripper device mountable to a user's hand includes a first finger rotatable inward by thumb force and a second finger rotatable inward by finger force, with a camera assembly mounted to capture image data of the fingers and workspace. The transmissionless design enables direct actuation by the user's digits, providing intuitive control and haptic feedback. The camera assembly captures both the manipulation workspace and surrounding environment to enable robust visual tracking. The device features integrated visual markers and compliant regions that provide grip position and force feedback without requiring sensors. The captured image data is used to train computational models that control robot manipulators having matching visual and mechanical properties.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for controlling a robot using language model programs. A language model program is computer program generated from an output of a code generation neural network, e.g., one that has been trained on a language modeling objective on computer code data.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for controlling an agent interacting with an environment. In one aspect, a method comprises: receiving one or more observations of an environment; receiving an input text sequence that describes a task to be performed by a robot in the environment; generating an encoded representation of the input text sequence in an embedding space; generating a corresponding encoded representation of each of the one or more observations in the embedding space; generating a sequence of input tokens that comprises the encoded representation of the input text sequence and the corresponding encoded representation of each observation; processing the sequence of input tokens using a language model neural network to generate an output text sequence that comprises high-level natural language instructions; and determining, from the high-level natural language instructions, one or more actions to be performed by the robot.

Abstract: There is described herein a method of prognosing or classifying a subject with a Myeloproliferative Neoplasm (MPN) comprising: (a) determining the expression level of at least 10 genes in a test sample from the subject selected from the group consisting of SPP1, CEACAM6, GJA1, IGSF10, IGFBP2, COL4A5, LYVE1, MTIE, EMP1, XIST, DLK1, TPSAB1, TIMP3, CLC, MS4A1, ENKUR, ALOX12, KNDC1, HLA-DQB1, GAS2, CLEC2L, BEND2, CDH7, and NT5E; and (b) comparing expression of the at least 10 genes in the test sample with reference expression levels of the at least 10 genes from control samples from a reference cohort of patients; wherein a difference or similarity in the expression of the at least 10 genes in the test sample and the reference expression levels is used to prognose or classify the subject with MPN into a low risk group or a high risk group for worse survival.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for controlling an agent interacting with an environment. In one aspect, a method comprises: receiving one or more observations of an environment; receiving an input text sequence that describes a task to be performed by a robot in the environment; generating an encoded representation of the input text sequence in an embedding space; generating a corresponding encoded representation of each of the one or more observations in the embedding space; generating a sequence of input tokens that comprises the encoded representation of the input text sequence and the corresponding encoded representation of each observation; processing the sequence of input tokens using a language model neural network to generate an output text sequence that comprises high-level natural language instructions; and determining, from the high-level natural language instructions, one or more actions to be performed by the robot.

Abstract: A transporter network for determining robot actions based on sensor feedback can afford robots with efficient autonomous movement. The transporter network may exploit spatial symmetries and may not need assumptions of objectness to provide accurate instructions on object manipulation. The machine-learned model of the transporter network may also allow for learning various tasks with less training examples than other machine-learned models. The machine-learned model of the transporter network may intake observation data as input and may output actions in response to the processed observation data.

Abstract: A transporter network for determining robot actions based on sensor feedback can afford robots with efficient autonomous movement. The transporter network may exploit spatial symmetries and may not need assumptions of objectness to provide accurate instructions on object manipulation. The machine-learned model of the transporter network may also allow for learning various tasks with less training examples than other machinelearned models. The machine-learned model of the transporter network may intake observation data as input and may output actions in response to the processed observation data.

Abstract: Systems and methods of the present disclosure are directed to computer-implemented method for contextual processing via inter-model between pre-trained machine-learned models. The method includes obtaining, by a computing system comprising one or more computing devices, input data. The method includes processing, by the computing system, the input data with two or more pre-trained models to generate output data, wherein processing the input comprises executing a structured inter-model communication schema for inter-model communication between the two or more pre-trained models over a communications channel. The method includes providing, by the computing system, the output data as an output.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for controlling an agent interacting with an environment. In one aspect, a method comprises: receiving one or more observations of an environment; receiving an input text sequence that describes a task to be performed by a robot in the environment; generating an encoded representation of the input text sequence in an embedding space; generating a corresponding encoded representation of each of the one or more observations in the embedding space; generating a sequence of input tokens that comprises the encoded representation of the input text sequence and the corresponding encoded representation of each observation; processing the sequence of input tokens using a language model neural network to generate an output text sequence that comprises high-level natural language instructions; and determining, from the high-level natural language instructions, one or more actions to be performed by the robot.

Abstract: A transporter network for determining robot actions based on sensor feedback can afford robots with efficient autonomous movement. The transporter network may exploit spatial symmetries and may not need assumptions of objectness to provide accurate instructions on object manipulation. The machine-learned model of the transporter network may also allow for learning various tasks with less training examples than other machinelearned models. The machine-learned model of the transporter network may intake observation data as input and may output actions in response to the processed observation data.

Abstract: A slider with a reboundable pull-tab includes a slider base, a pull-tab, a connecting shaft and spring; the slider base has a connecting shaft through-hole for mounting the connecting shaft, which fixes the pull-tab and spring onto the slider base; one end of the spring inserts into the slider, and the other end inserts into the spring base; when the pull-tab is pulled, the spring will subject to torsional deformation; when the pull-tab is let go, the pull-tab will revert to the normal state under the impetus of the spring, then the slider is automatically locked, thus preventing unexpectedly sliding; there is also a locking part in the pull-tab, so that it is free from not being able to rebound due to over-rotation.

Abstract: The present invention discloses a slider with an reboundable pull-tab, comprising a slider base, a pull-tab, a connecting shaft and spring; the slider base has a connecting shaft through-hole for mounting the connecting shaft, which fixes the pull-tab and spring onto the slider base; one end of the spring inserts into the slider, and the other end inserts into the spring base; when the pull-tab is pulled, the spring will subject to torsional deformation; when the pull-tab is let go, the pull-tab will revert to the normal state under the impetus of the spring, then the slider is automatically locked, thus preventing unexpectedly sliding; There is also a locking part in the pull-tab, so that it is free from not being able to rebound due to over-rotation.