Abstract: A method includes obtaining a rule that is associated with a segment of a map, determining a state for use in evaluation of the rule, evaluating the rule using the state, identifying an action to be performed based on a result of evaluating the rule using the state, and utilizing the action to be performed as an input to an automated control system of a vehicle.

Abstract: Systems and methods for moving point detection for point clouds. Some implementations may include obtaining a sequence of point clouds, wherein the sequence of point clouds includes a current point cloud, a previous point cloud that precedes the current point cloud, and a next point cloud that follows the current point cloud; inputting a voxelized representation of the current point cloud to a forward-pass recurrent neural network with a hidden state based on the previous point cloud to obtain a first voxelized activation; inputting the voxelized representation of the current point cloud to a backward-pass recurrent neural network with a hidden state based on the next point cloud to obtain a second voxelized activation; combining the first voxelized activation and the second voxelized activation to obtain a prediction of whether respective points are moving or static; and updating the current point cloud based on the prediction.

Abstract: Systems and methods for semantic labeling of point clouds using images. Some implementations may include obtaining a point cloud that is based on lidar data reflecting one or more objects in a space; obtaining an image that includes a view of at least one of the one or more objects in the space; determining a projection of points from the point cloud onto the image; generating, using the projection, an augmented image that includes one or more channels of data from the point cloud and one or more channels of data from the image; inputting the augmented image to a two dimensional convolutional neural network to obtain a semantic labeled image wherein elements of the semantic labeled image include respective predictions; and mapping, by reversing the projection, predictions of the semantic labeled image to respective points of the point cloud to obtain a semantic labeled point cloud.

Abstract: A method includes capturing images at a device and analyzing the images at the device using a first analysis model to obtain information regarding an observed object that corresponds to a predetermined object. The method also includes determining whether the information regarding the observed object is consistent with stored mapping information, and, in response to determining that the information regarding the observed object is not consistent with the stored mapping information, modifying the stored mapping information based on the observed object.

Abstract: A method includes obtaining a three-dimensional model that represents a natural environment and identifying ground-level surfaces from the three-dimensional model, wherein the ground-level surfaces from the three-dimensional model represent ground-level features from the natural environment. The method also includes identifying overhanging features from the three-dimensional model, wherein the overhanging features from the three-dimensional model block visibility of the ground-level surfaces from the three-dimensional model when the ground-level surfaces are viewed from an orthographic perspective.

Abstract: A method includes capturing images at a device and analyzing the images at the device using a first analysis model to obtain information regarding an observed object that corresponds to a predetermined object. The method also includes determining whether the information regarding the observed object is consistent with stored mapping information, and, in response to determining that the information regarding the observed object is not consistent with the stored mapping information, modifying the stored mapping information based on the observed object.

Abstract: Systems and methods for semantic labeling of point clouds using images. Some implementations may include obtaining a point cloud that is based on lidar data reflecting one or more objects in a space; obtaining an image that includes a view of at least one of the one or more objects in the space; determining a projection of points from the point cloud onto the image; generating, using the projection, an augmented image that includes one or more channels of data from the point cloud and one or more channels of data from the image; inputting the augmented image to a two dimensional convolutional neural network to obtain a semantic labeled image wherein elements of the semantic labeled image include respective predictions; and mapping, by reversing the projection, predictions of the semantic labeled image to respective points of the point cloud to obtain a semantic labeled point cloud.

Abstract: A method includes obtaining a three-dimensional model that represents a natural environment and identifying ground-level surfaces from the three-dimensional model, wherein the ground-level surfaces from the three-dimensional mode represent ground-level features from the natural environment. The method also includes identifying overhanging features from the three-dimensional model, wherein the overhanging features from the three-dimensional model block visibility of the ground-level surfaces from the three-dimensional model when the ground-level surfaces are viewed from an orthographic perspective.

Abstract: Systems and methods for moving point detection for point clouds. Some implementations may include obtaining a sequence of point clouds, wherein the sequence of point clouds includes a current point cloud, a previous point cloud that precedes the current point cloud, and a next point cloud that follows the current point cloud; inputting a voxelized representation of the current point cloud to a forward-pass recurrent neural network with a hidden state based on the previous point cloud to obtain a first voxelized activation; inputting the voxelized representation of the current point cloud to a backward-pass recurrent neural network with a hidden state based on the next point cloud to obtain a second voxelized activation; combining the first voxelized activation and the second voxelized activation to obtain a prediction of whether respective points are moving or static; and updating the current point cloud based on the prediction.

Abstract: A method includes obtaining images, obtaining three-dimensional surface measurements, defining a three-dimensional mesh using the three-dimensional surface measurements, texturing the three-dimensional mesh using the images to define a textured three-dimensional mesh, identifying a first portion of the textured three-dimensional mesh, identifying a second portion of the textured three-dimensional mesh that obstructs visibility of part of the first portion of the textured three-dimensional mesh from an overhead perspective, and rendering a simulated overhead perspective image such that the second portion of the textured three-dimensional mesh is not represented in the simulated overhead perspective image.

Abstract: A method includes obtaining images, obtaining three-dimensional surface measurements, defining a three-dimensional mesh using the three-dimensional surface measurements, texturing the three-dimensional mesh using the images to define a textured three-dimensional mesh, identifying a first portion of the textured three-dimensional mesh, identifying a second portion of the textured three-dimensional mesh that obstructs visibility of part of the first portion of the textured three-dimensional mesh from an overhead perspective, and rendering a simulated overhead perspective image such that the second portion of the textured three-dimensional mesh is not represented in the simulated overhead perspective image.

Abstract: Systems and methods for semantic labeling of point clouds using images. Some implementations may include obtaining a point cloud that is based on lidar data reflecting one or more objects in a space; obtaining an image that includes a view of at least one of the one or more objects in the space; determining a projection of points from the point cloud onto the image; generating, using the projection, an augmented image that includes one or more channels of data from the point cloud and one or more channels of data from the image; inputting the augmented image to a two dimensional convolutional neural network to obtain a semantic labeled image wherein elements of the semantic labeled image include respective predictions; and mapping, by reversing the projection, predictions of the semantic labeled image to respective points of the point cloud to obtain a semantic labeled point cloud.