Data segmentation using masks
AssigneeZoox, Inc.
InventorsZeng Wang, David Pfeiffer, Dragomir Dimitrov Anguelov, Subhasis Das, Allan Zelener
Life of this patent
Estimating the term…
Abstract
A vehicle can include various sensors to detect objects in an environment. Sensor data can be captured by a perception system in a vehicle and represented in a voxel space. Operations may include analyzing the data from a top-down perspective. From this perspective, techniques can associate and generate masks that represent objects in the voxel space. Through manipulation of the regions of the masks, the sensor data and/or voxels associated with the masks can be clustered or otherwise grouped to segment data associated with the objects.
Claims20
A system comprising:
- one or more processors;
- and
- one or more computer-readable media storing instructions executable by the one or more processors, wherein the instructions, when executed, cause the system to perform operations comprising:
- capturing sensor data using a light detection and ranging (LIDAR) sensor on a vehicle;
- associating the sensor data with a voxel space, the voxel space comprising at least three dimensions;
- generating a two-dimensional representation of the voxel space, the two-dimensional representation comprising a number of image channels;
- inputting the two-dimensional representation into a machine learning algorithm;
- receiving, from the machine learning algorithm, a first mask representing an object in the voxel space in two dimensions;
- generating, based at least in part on an expansion of the first mask, a second mask, the expansion based at least in part on a region growing algorithm, a size of the first mask, or an intersection with a third mask associated with another object;
- and
- segmenting, based at least in part on the second mask, the sensor data.
A method comprising:
- capturing sensor data of an environment using one or more sensors, the sensor data indicative of an object in the environment;
- associating the sensor data with a voxel space;
- receiving a first mask associated with a portion of the voxel space, the first mask representing a region smaller in size than the object;
- generating a second mask by expanding the first mask;
- and
- segmenting, based at least in part on the second mask, the sensor data.
A non-transitory computer-readable medium storing instructions executable by one or more processors, wherein the instructions, when executed, cause the one or more processors to perform operations comprising:
- receiving sensor data of an environment from one or more sensors on an autonomous vehicle, the sensor data associated with an object in the environment;
- associating the sensor data with a voxel space;
- receiving a first mask associated with a portion of the voxel space, the first mask representing at least a portion of the object using a second perspective;
- generating a second mask by expanding the first mask;
- and
- segmenting, based at least in part on the second mask, the sensor data.
Dependencies are inferred from the claim text.