Abstract: Provided are systems and methods for synthesizing novel views of complex scenes (e.g., outdoor scenes). In some implementations, the systems and methods can include or use machine-learned models that are capable of learning from unstructured and/or unconstrained collections of imagery such as, for example, “in the wild” photographs. In particular, example implementations of the present disclosure can learn a volumetric scene density and radiance represented by a machine-learned model such as one or more multilayer perceptrons (MLPs).

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating object interaction predictions using a neural network. One of the methods includes obtaining a sensor input derived from data generated by one or more sensors that characterizes a scene. The sensor input is provided to an object interaction neural network. The object interaction neural network is configured to process the sensor input to generate a plurality of object interaction outputs. Each respective object interaction output includes main object information and interacting object information. The respective object interaction outputs corresponding to the plurality of regions in the sensor input are received as output of the object interaction neural network.

Abstract: Provided are systems and methods for synthesizing novel views of complex scenes (e.g., outdoor scenes). In some implementations, the systems and methods can include or use machine-learned models that are capable of learning from unstructured and/or unconstrained collections of imagery such as, for example, “in the wild” photographs. In particular, example implementations of the present disclosure can learn a volumetric scene density and radiance represented by a machine-learned model such as one or more multilayer perceptrons (MLPs).

Abstract: Provided are systems and methods for synthesizing novel views of complex scenes (e.g., outdoor scenes). In some implementations, the systems and methods can include or use machine-learned models that are capable of learning from unstructured and/or unconstrained collections of imagery such as, for example, “in the wild” photographs. In particular, example implementations of the present disclosure can learn a volumetric scene density and radiance represented by a machine-learned model such as one or more multilayer perceptrons (MLPs).

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating object interaction predictions using a neural network. One of the methods includes obtaining a sensor input derived from data generated by one or more sensors that characterizes a scene. The sensor input is provided to an object interaction neural network. The object interaction neural network is configured to process the sensor input to generate a plurality of object interaction outputs. Each respective object interaction output includes main object information and interacting object information. The respective object interaction outputs corresponding to the plurality of regions in the sensor input are received as output of the object interaction neural network.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating object interaction predictions using a neural network. One of the methods includes obtaining a sensor input derived from data generated by one or more sensors that characterizes a scene. The sensor input is provided to an object interaction neural network. The object interaction neural network is configured to process the sensor input to generate a plurality of object interaction outputs. Each respective object interaction output includes main object information and interacting object information. The respective object interaction outputs corresponding to the plurality of regions in the sensor input are received as output of the object interaction neural network.

Abstract: Methods, systems, and apparatus, including computer programs encoded on a computer storage medium, for generating object interaction predictions using a neural network. One of the methods includes obtaining a sensor input derived from data generated by one or more sensors that characterizes a scene. The sensor input is provided to an object interaction neural network. The object interaction neural network is configured to process the sensor input to generate a plurality of object interaction outputs. Each respective object interaction output includes main object information and interacting object information. The respective object interaction outputs corresponding to the plurality of regions in the sensor input are received as output of the object interaction neural network.

Abstract: A method for starting an operating system includes transferring, in response to starting an operating system in a data processing system, a first memory image that corresponds to a bootable storage medium to a working memory of the data processing system during a transfer of data. The method also includes processing, by a processor of the data processing system, instructions associated with the first memory image during a command execution process. The method also includes starting, by the processor, the command execution process for the instructions associated with the first memory image before transfer of the first memory image by the data transfer to the working memory of the data processing system is completed.

Abstract: A dynamic map display presents a dynamic image on a display screen of a vehicle. The dynamic map display includes a camera aimed at and above the horizon in front of the vehicle to capture a background image, which is transmitted to a controller via a first communications channel. A map generator within the controller generates a map image showing navigational directions. The controller also includes an image generator combining the background image with the map image as the dynamic image, which is transmitted to the display screen via a second communications channel. The background image and the map image are each complementary shaped to fill the display screen in any one of several different views and with the background image and the map image having different shapes and/or sizes. A method for generating a dynamic image is also disclosed.

Abstract: A method for processing and presenting information to a vehicle occupant via a vehicle interface system is provided. The method includes presenting information to the vehicle occupant via at least one electronic display of the vehicle interface system and running a plurality of software applications on the vehicle interface system. The method further includes connecting a user device to the vehicle interface system and identifying a non-vehicle-specific version of one of the plurality of software applications installed on the user device. The method further includes installing a vehicle-specific version of the identified software application on the vehicle interface system in response to identifying the non-vehicle-specific version of the software application installed on the user device.

Abstract: A vehicle interface system includes a graphics processing unit and a plurality of processing domains. The processing domains execute vehicle applications and generate tasks for the graphics processing unit. The system further includes a task scheduler configured to receive the tasks generated by the processing domains and to determine an order in which to send the tasks to the graphics processing unit. The graphics processing unit processes the tasks in the order determined by the task scheduler and generates display data based on the tasks. The system further includes an electronic display configured to receive the display data generated by the graphics processing unit and to present the display data to a user.

Abstract: A vehicle interface system includes at least one electronic display configured to present information to an occupant of the vehicle. The system includes a processing circuit including one or more processors and memory. The memory stores a plurality of software applications configured to provide display data for presentation via the electronic display. The processing circuit includes a display mapper configured to generate a plurality of adjustable virtual operating fields and to assign each of the virtual operating fields to one of the software applications. Each of the virtual operating fields covers a non-overlapping portion of the electronic display and presents the display data provided by the software application to which the virtual operating field is assigned.

Abstract: A computer system for integration with a vehicle user interface includes a processing system. The processing system includes a multi-core processor. The processing system is configured to provide virtualization for a first guest operating system in a first core or cores of the multi-core processor. The processing system is also configured to provide virtualization for a second guest operating system in a second and different core or cores of the multi-core processor. The first guest operating system is configured for high reliability operation. The virtualization prevents operations of the second guest operating system from disrupting the high reliability operation of the first guest operating system.

Abstract: A process is provided for forming a rapid-hardening inorganic foam, based on the reaction of the following two components: (a) a solid component in the form of a reactive powder having at the same time structure-building and pore-forming properties, and (b) a liquid component in the form of an alkali metal silicate (water glass). The reactive powder contains the essential constituents: 45-65% by weight aluminum oxide, 10-20% by weight aluminum nitride, and 5-15% by weight metallic aluminum. The alkali metal silicate has a molar ratio of silicon oxide to metal oxide of 1.0 to 2.2. The two components are mixed in a weight ratio of powder to liquid component of 0.5 to 2 to give a paste from which a foamed body having a bulk density of less than 0.7 g/cm3 is then formed in an exothermal reaction within less than 10 minutes.