University of Minnesota

Google Inc.

Chao X Guo

Stergios I Roumeliotis

Luca Ballan

Qiyue Zhang

Mahesh Ramachandran

Joel Hesch

Georgios Georgiou

Konstantine Nicholas John Tsotsos

Kejian J Wu

Hongsheng Yu

Fengtao Fan

Jeffrey Neil Margolis

Scott Fullam

Esha D Nerurkar

Eric Turner

Yunwen Zhou

Dimitrios G Kottas

Zhiheng Jia

Abhishek Kar

Sazzadur Rahman

Ryan Christopher DuToit

Jie Zhang

Ryan C DuToit

Fei Han

Kourosh Sartipi

Junyang Lu

A vision-aided inertial navigation system (VINS) comprises an image source for producing image data along a trajectory. The VINS further comprises an inertial measurement unit (IMU) configured to produce IMU data indicative of motion of the VINS and an odometry unit configured to produce odometry data. The VINS further comprises a processor configured to compute, based on the image data, the IMU data, and the odometry data, state estimates for a position and orientation of the VINS for poses of the VINS along the trajectory. The processor maintains a state vector having states for a position and orientation of the VINS and positions within the environment for observed features for a sliding window of poses. The processor applies a sliding window filter to compute, based on the odometry data, constraints between the poses within the sliding window and compute, based on the constraints, the state estimates.

Vision-aided inertial navigation system for ground vehicle localization

Vision-aided inertial navigation techniques are described. In one example, a vision-aided inertial navigation system (VINS) comprises an image source to produce image data at a first set of time instances along a trajectory within a three-dimensional (3D) environment, wherein the image data captures features within the 3D environment at each of the first time instances. An inertial measurement unit (IMU) to produce IMU data for the VINS along the trajectory at a second set of time instances that is misaligned with the first set of time instances, wherein the IMU data indicates a motion of the VINS along the trajectory. A processing unit comprising an estimator that processes the IMU data and the image data to compute state estimates for 3D poses of the IMU at each of the first set of time instances and 3D poses of the image source at each of the second set of time instances along the trajectory. The estimator computes each of the poses for the image source as a linear interpolation from a subset of the poses for the IMU along the trajectory.

Efficient vision-aided inertial navigation using a rolling-shutter camera with inaccurate timestamps

Techniques include performing a calibration of the display with respect to the eye-tracking camera after usage. The calibration involves determining a current position of a reflected pixel of the display within a field of view of the eye-tracking camera and comparing that position with a calibration position, for example a position seen at the factory. The pixel may then be shifted according to the difference between the current position and the calibration position.

Extrinsic calibration between display and eye-tracking camera in head-mounted devices

Disclosed is a method including receiving a depth map estimated using data based on image and data received from a movement sensor as input, generating an alignment parameter based on the depth map, adding the alignment parameter to a pre-calibration state to define a user operational calibration state, generating scale parameters and shift parameters based on features associated with the data received from the image and movement sensor, and calibrating the image and movement sensor based on the user operational calibration state, the scale parameters and the shift parameters.

Visual inertial odometry with machine learning depth

A tracking system and associated algorithms are disclosed that can provide a user with a continuous, reset-free augmented reality (AR) experience. When the user wears an AR headset equipped with a camera and an inertial measurement unit (IMU), motion sensor data from the IMU can be combined with image data from the camera to create a device pose, representing a position and an orientation of the headset relative to its environment. In some implementations, when a reset occurs, a six-degrees-of-freedom (6DoF) algorithm can be configured to support the pose until a re-initialization is completed. In some implementations, a neural network can be used to correct for IMU integration drifts in the 6DoF algorithm. In some implementations, the IMU-based 6DoF uses a neural network that exploits the device's past motion to infer its future motion.

Pose algorithm for continuous AR experiences

Techniques of providing continual augmented reality navigation include combining first and second 6DoF poses from, respectively, a localization from a map in a vicinity of a location to an image and from predicted 6DoF poses based on IMU data. The first 6DoF pose is based on a visual positioning system that uses map data from a vicinity of a location combined with an image from the world-facing camera of the AR smartglasses that provides 6DoF pose in world coordinates independent of the IMU. The second 6DoF pose is based on compensated rotational velocity and acceleration measurements derived from IMU intrinsic values.

Map-aided inertial odometry with neural network for augmented reality devices

Systems and techniques for persistent calibration of an electronic device configured to implement an extended reality (XR) system involve estimating and validating visual-inertial odometry (VIO) calibration parameters during an active XR session of the electronic device. Validating the estimated VIO calibration parameters involves performing a strict calibration qualification of the estimated VIO calibration parameters using a thresholding module, machine learning module, or both. An initial calibration qualification is performed concurrently with the strict calibration qualification based on VIO performance. If the estimated VIO calibration parameters pass the strict calibration qualification and initial calibration qualification, they are stored for use to calibrate the device in future XR sessions. Persistent calibration of time alignment between the inertial management unit and the image sensor of the electronic device is also performed during active XR sessions upon detection of time alignment issues.

Persistent calibration of extended reality systems

Techniques of providing continual augmented reality navigation include combining first and second 6DoF poses from, respectively, a localization from a map in a vicinity of a location to an image and from predicted 6DoF poses based on IMU data.

Efficient vision-aided inertial navigation using a rolling-shutter camera

A method includes: receiving, with a computing platform, respective trajectory data and map data independently generated by each of a plurality of vision-aided inertial navigation devices (VINS devices) traversing an environment, wherein the trajectory data specifies poses along a path through the environment for the respective VINS device and the map data specifies positions of observed features within the environment as determined by an estimator executed by the respective VINS device; determining, with the computing platform and based on the respective trajectory data and map data from each of the VINS devices, estimates for relative poses within the environment by determining transformations that geometrically relate the trajectory data and the map data between one or more pairs of the VINS devices; and generating, with the computing platform and based on the transformations, a composite map specifying positions within the environment for the features observed by the VINS devices.

Resource-aware large-scale cooperative 3D mapping using multiple mobile devices

This disclosure describes techniques for reducing or eliminating estimator inconsistency in vision-aided inertial navigation systems (VINS). For example, an observability-constrained VINS (OC-VINS) is described which enforce the unobservable directions of the system to prevent spurious information gain and reduce inconsistency.

Growing community of inventors

San Jose, CA, United States of America

Chao X Guo