Amazon Technologies, Inc.

Argo Ai Gmbh

Volkswagen Group of America Investments, LLC

Ankit Kumar Jain

Gang Pan

Murtaza Muidul Huda Chowdhury

Ramapulla Reddy Chennuru

Xxx Xinjilefu

Niti S Khadapkar

Anupama Anand

Weizhao Shao

Brendan Christopher Byrne

Bhaven Avalani

Alexey Gadalin

Andi Gutmans

Srinivasan Ramkumar

Anup P Pandya

Min Zhou

John Edsel Santos

Ramakrishna Kotla

Ankit Malpani

Siddharth Padmanabhan Ramanarayanan

Jaco Joubert

Xiangyu Ren

Wenjian Guo

Yashi Kamboj

Eli R Fisher

Systems and techniques are described herein for tiered storage of customer data accessed by a search service of a computing resource service provider. In some aspects, customer data may be received by a search instance executed across a plurality of compute nodes and provisioned by a search service. The customer data may be indexed and the data and resulting index may be stored locally by a first pool of hot compute nodes of the search instance. The customer data and index may be migrated and stored remotely by a data storage service. Metadata associated with the customer data and/or index may be stored in a second pool of warm compute nodes of the search instance. The warm compute nodes, upon receiving a request to access the customer data, may identify a location of the customer data and retrieve the customer data from the data storage service according to the metadata.

Warm tier storage for search service

A system and method for performing visual odometry is disclosed. In aspects, the system implements methods to generate an image pyramid based on an input image received. A refined pose prior information representing a location and orientation of the autonomous vehicle can be generated based on one or more images of the image pyramid. One or more seed points can be selected from the one or more images of the image pyramid. One or more refined seed representing the one or more seed points with added depth values can be generated. One or more scene points can be generated based on the one or more refined seed points. A point cloud can be generated based on the one or more scene points.

Heterogeneous multi-threaded visual odometry in autonomous driving

A system and method for performing visual localization is disclosed. In aspects, the system implements methods to generate a global point cloud, the global point cloud representing a plurality of point clouds. The global point cloud can be mapped to a prior map information to locate a position of an autonomous vehicle, the prior map information representing pre-built geographic maps. The position of the autonomous vehicle can be estimated based on applying sensor information obtained from sensors and software of the autonomous vehicle to the mapped global point cloud.

Visual localization against a prior map

Techniques for host resource group sharing in a provider network are described. A resource share is made of a host resource group made up of dedicated hosts in the provider network. The share provides access to the host resource group of a primary account with one or more other principals—e.g., user accounts, an organization, or a group of users. The other principals can launch compute instances into the host resource group, which may cause the automatic provisioning of new hosts into the host resource group, of the primary account, as needed.

Shared dedicated host resource groups in a provider network

Techniques for automated defragmentation of licensed resources in a provider network are described. A defragmentation service in a provider network can detect the existence of intra-host fragmentation and/or inter-host fragmentation among host computing devices belonging to a host resource group of hosts dedicated to particular users. The defragmentation service can cause instances to be migrated to new locations within the host resource group to reduce or eliminate the fragmentation, allowing for hosts to beneficially be removed from the host resource group. The defragmentation service may determine which instances to be moved based on considerations involving software licenses associated with the host computing devices.

Defragmentation of licensed resources in a provider network

This disclosure describes techniques for flexible maintenance windows for performing maintenance for instances. Using techniques described herein, a user of a service provider network may specify configuration data that relates to maintenance events of instances. A 'maintenance event' is any event that relates to deploying instances and/or at least temporarily losing use of an instance. For example, a maintenance event may cause an instance to be deployed, re-booted, re-started, or replaced. According to some configurations, customers may specify one or more maintenance event windows for when maintenance is to be performed, as well as customize notifications that are provided to the customer about scheduled maintenance events. For instance, customers may create a maintenance window that specifies to perform maintenance on Tuesdays, between 2 AM-3 AM, and that notifications are to include information specified by the customer (e.g., through one or more tags).

Utilizing maintenance event windows to determine placement of instances

Growing community of inventors

Mountain View, CA, United States of America

Ankit Kumar Jain