Google Inc.

Deepmap Inc.

Aurora Operations, Inc.

Matterport, Inc.

Nvidia Corporation

Greg Coombe

Mark Damon Wheeler

Chen Chen

Jeffrey Minoru Adachi

Francois Bailly

Daniel Barcay

Gokul Varadhan

Derik Schroeter

Yekeun Jeong

Dorra Larnaout

Systems, computer-implemented methods, apparatus and/or computer program products are provided that facilitate improving the accuracy of global positioning system (GPS) coordinates of indoor photos. The disclosed subject matter further provides systems, computer-implemented methods, apparatus and/or computer program products that facilitate generating exterior photos of structures based on GPS coordinates of indoor photos.

Accuracy of GPS coordinates associated with image capture locations

Map data describing various geometric elements in a map used in autonomous vehicle localization is selectively weighted to vary the relative contributions of different geometric elements to an alignment operation. By doing so, the alignment operation may be biased to emphasize geometric elements associated with objects in an environment that are relatively more stable, e.g., roadways, walls, buildings, etc., over objects that are relatively less stable, e.g., vegetation, thereby in many instances improving alignment operation performance and/or availability.

Generation of weighted map data for autonomous vehicle localization

A high-definition map system receives sensor data from vehicles travelling along routes and combines the data to generate a high definition map for use in driving vehicles, for example, for guiding autonomous vehicles. A pose graph is built from the collected data, each pose representing location and orientation of a vehicle. The pose graph is optimized to minimize constraints between poses. Points associated with surface are assigned a confidence measure determined using a measure of hardness/softness of the surface. A machine-learning-based result filter detects bad alignment results and prevents them from being entered in the subsequent global pose optimization. The alignment framework is parallelizable for execution using a parallel/distributed architecture. Alignment hot spots are detected for further verification and improvement. The system supports incremental updates, thereby allowing refinements of subgraphs for incrementally improving the high-definition map for keeping it up to date.

Classification of surfaces as hard/soft for combining data captured by autonomous vehicles for generating high definition maps

Distributed processing of pose graphs for generating high definition maps for navigating autonomous vehicles

A staged camera traversal for navigating a virtual camera in a three dimensional environment is provided. The staged camera traversal can include a launch stage and an approach stage. During the launch stage, the tilt angle can be decreased towards zero tilt (i.e. straight down) with respect to the vertical. During an approach stage, the tilt angle of the virtual camera can be increased from about zero tilt towards the tilt angle associated with a target location. In certain implementations, the staged camera traversal can further include a traversal stage occurring between the launch stage and the approach stage. The tilt angle of the virtual camera can be maintained at about zero tilt during the traversal stage. The approach path of the virtual camera can be aligned along a view direction associated with the target destination during the approach stage.

Staged camera traversal for three dimensional environment

Embodiments alter the swoop trajectory depending on the terrain within the view of the virtual camera. To swoop into a target, a virtual camera may be positioned at an angle relative to the upward normal vector from the target. That angle may be referred to as a tilt angle. According to embodiments, the tilt angle may increase more quickly in areas of high terrain variance (e.g., mountains or cities with tall buildings) than in areas with less terrain variance (e.g., flat plains). To determine the level of terrain variance in an area, embodiments may weigh terrain data having higher detail more heavily than terrain data having less detail.

Terrain-based virtual camera tilting, and applications thereof

The capability to render and navigate three-dimensional panoramic images in a virtual three-dimensional environment so as to create an immersive three-dimensional experience is provided. Such a capability can present a three-dimensional photographic experience of the real world that is seamlessly integrated with the virtual three-dimensional environment. Depth values associated with the panoramic images may be used to create three-dimensional geometry, which can be rendered as part of the virtual three-dimensional environment. Further, such a capability can enable a user to roam freely through the environment while providing a more natural free-form exploration of the environment than existing systems.

Growing community of inventors

Mountain View, CA, United States of America

Greg Coombe