LiDAR localization using 3D CNN network for solution inference in autonomous driving vehicles

A system for extracting point cloud features for solution inference using neural networks in light detection and range (LiDAR) localization, the system comprising:

• a processor;
• and
• a memory coupled to the processor to store instructions, which when executed by the processor, cause the processor to perform operations, the operations including:
• constructing a cost volume in a solution space for a predicted pose of an autonomous driving vehicle (ADV), wherein the cost volume includes a plurality of sub volumes, each sub volume representing a matching cost between a keypoint from an online point cloud and a corresponding keypoint on a pre-built point cloud map, and wherein the cost volume is constructed based on pairs of corresponding feature descriptors extracted from the online point cloud and the pre-built point cloud map,
• regularizing the cost volume using a plurality of convention neural networks (CNNs) to refine the matching costs, and
• inferring, from the regularized cost volume, an optimal offset of the predicted pose, the optimal offset used to determine a location of the ADV.

A non-transitory machine-readable medium having instructions stored therein, which when executed by a processor, cause the processor to perform operations for extracting point cloud features for solution inference using neural networks in light detection and range (LiDAR) localization, the operations comprising:

• constructing a cost volume in a solution space for a predicted pose of an autonomous driving vehicle (ADV), wherein the cost volume includes a plurality of sub volumes, each sub volume representing a matching cost between a keypoint from an online point cloud and a corresponding keypoint on a pre-built point cloud map, and wherein the cost volume is constructed based on pairs of corresponding feature descriptors extracted from the online point cloud and the pre-built point cloud map;
• regularizing the cost volume using a plurality of convention neural networks (CNNs) to refine the matching costs;
• and
• inferring, from the regularized cost volume, an optimal offset of the predicted pose, the optimal offset used to determine a location of the ADV.