Toyota Research Institute, Inc.

Embark Trucks Inc.

Konstantine Mushegian

Kevin Stone

Brandon Northcutt

Katarina Bouma

Matthew W Amacker

Liam Benson

Hourann William Bosci

Paul Ashbourne

Caleb William Locke Foust

Akiyoshi Ochiai

Kazuto Murase

A method for training an object detection system includes estimating a location of a first object in an environment based on a density cluster map generated from a plurality of images of the environment. The method also includes generating one or more negative training samples of the first object in the environment based on the plurality of images, each of the one or more negative training samples corresponding to a second object at a location in the environment that is different than the estimated location of the first object. The method further includes generating positive training samples from a set of images of the first object. The method also includes training the object detection system to detect the first object based on the positive training samples and the negative training sample.

Object detection improvement based on autonomously selected training samples

A system and method including receiving an indication of an occurrence of an event associated with the operation of an autonomous vehicle; determining at least one action to be performed, the at least one action including a data request for a specified subset of stored data associated with the operation of the autonomous vehicle from a memory; generating an output including the specified subset of data; and transmitting the specified subset of data to a remote monitoring system.

Infrastructure to support real-time review of events

Provided are systems, methods and computer program code for transmitting vehicle data to remote monitoring systems using a low bandwidth protocol.

Low bandwidth protocol for streaming sensor data

A method for generating positive and negative training samples is presented. The method includes identifying false positive images of an object based on multiple images of an environment. The method also includes generating positive training samples from a set of images of the object. The method further includes generating a negative training sample from the false positive image. The method still further includes training an object detection system based on the positive training samples and the negative training sample.

A system for remote visual inspection of a closed space includes a base station including a distance finder and a light emitter. The base station determines a distance to a projection surface using the distance finder and projects a pattern onto the projection surface of the closed space at a projection location on the projection surface. A robot includes a moveable base supporting an imaging device, a processor, and a storage device storing one or more non-transitory, processor-readable instructions. When executed by the processor, the instructions cause the robot to detect the pattern with the imaging device, determine a location of the robot with respect to the base station based on the pattern, and capture image data of the closed space. An external electronic device is communicatively coupled to the robot. The external electronic device receives image data and displays one or more images based on the image data.

Systems and methods for remote visual inspection of a closed space

An orientation system includes a base station comprising a light emitter that projects a pattern onto a projection surface of a space, an auxiliary station including an auxiliary emitter that projects an auxiliary pattern onto the projection surface, and a robot including a moveable base supporting an imaging device. The robot detects the pattern and the auxiliary pattern with the imaging device, and determines a location of the robot within the space based on the pattern.

Systems and methods for orienting a robot in a space

Systems, robots, and methods for generating three-dimensional skeleton representations of people are disclosed. A method includes generating, from a two-dimensional image, a two-dimensional skeleton representation of a person present in the two-dimensional image. The two-dimensional skeleton representation includes a plurality of joints and a plurality of links between individual joints of the plurality of joints. The method further includes positioning a cone around one or more links of the plurality of links, and identifying points of a depth cloud that intersect with the cone positioned around the one or more links of the two-dimensional skeleton. The points of the depth cloud are generated by a depth sensor and each point provides depth information. The method also includes projecting the two-dimensional skeleton representation into three-dimensional space using the depth information of the points that intersect with the cone, thereby generating the three-dimensional skeleton representation of the person.

Growing community of inventors

San Francisco, CA, United States of America

Konstantine Mushegian