Waymo LLC

Waymo L Lc

Mingcheng Chen

Pierre-yves Droz

Pierre-Yves Droz

Christian Lauterbach

Xiaoxiang Hu

Colin Andrew Braley

Nicholas Armstrong-Crews

Benjamin Pitzer

Craig Lewin Robinson

William Grossman

Peter Morton

Yunshan Jiang

Blaise Gassend

Mark Alexander Shand

Luke Wachter

Volker Grabe

Gil Shotan

Nathaniel Quillin

Richard Chaussee

Sachithra Hemachandra

The technology relates to tracking objects in an environment around an autonomous vehicle. A computing system of the autonomous vehicle determines accurate motion characteristics of objects detected in its environment despite various sensor measurement limitations. By correcting motion distortion for fast moving objects and accounting for discrepancies in sensor data gathering, motion characteristics may be determined for the detected objects with enhanced accuracy. Multiple sets of correspondences are determined for clusters from multiple sensor spins, enabling better alignment using a surface matching algorithm even when clusters have fewer data points. Efficiency is also enhanced by selecting hypotheses based on confidence levels. These techniques provide for identifying the types of objects for which a yaw rate can be accurately determined. Object classification can also be improved by accumulating associated clusters corresponding to a detected object. In addition, under- or over-segmentation can be mitigated with such techniques.

Onboard cluster tracking system

The technology relates to autonomous vehicles having articulating sections such as the trailer of a tractor-trailer. Aspects include approaches for tracking the pose of the trailer, including its orientation relative to the tractor unit. Sensor data is analyzed from one or more onboard sensors to identify and track the pose. The pose information is usable by on-board perception and/or planning systems when driving the vehicle in an autonomous mode. By way of example, on-board sensors such as Lidar sensors are used to detect the real-time pose of the trailer based on Lidar point cloud data. The orientation of the trailer is estimated based on the point cloud data, and the pose is determined according to the orientation and other information about the trailer. Aspects also include determining which side of the trailer the sensor data is coming from. A camera may also detect trailer marking information to supplement the analysis.

Lidar-based trailer tracking

Aspects and implementations of the present disclosure address shortcomings of the existing technology by enabling Doppler-assisted segmentation of points in a point cloud for efficient object identification and tracking in autonomous vehicle (AV) applications, by: obtaining, by a sensing system of the AV, a plurality of return points comprising one or more velocity values and one or more coordinates of a reflecting region that reflects a signal emitted by the sensing system, the one or more velocity values and the one or more coordinates obtained for the same instance of time, identifying that the set of the return points is associated with an object in an environment, and causing a driving path of the AV to be determined in view of the object.

Point cloud segmentation using a coherent lidar for autonomous vehicle applications

Aspects and implementations of the present disclosure address shortcomings of the existing technology by enabling motion pattern-assisted object classification of objects in an environment of an autonomous vehicle (AV) by obtaining, from a sensing system of the AV, a plurality of return points, each return point comprising one or more velocity values and one or more coordinates of a reflecting region that reflects a signal emitted by the sensing system, identifying an association of the plurality of return points with an object in an environment of the AV, identifying, in view of the one or more velocity values of at least some of the plurality of return points, a type of the object or a type of a motion of the object, and causing a driving path of the AV to be determined in view of the identified type of the object.

Classification of objects based on motion patterns for autonomous vehicle applications

Aspects and implementations of the present disclosure address shortcomings of the existing technology by enabling velocity estimation for efficient object identification and tracking in autonomous vehicle (AV) applications, including: obtaining, by a sensing system of the AV, a plurality of return points, each return point having a velocity value and coordinates of a reflecting region that reflects a signal emitted by the sensing system, identifying an association of the velocity values and the coordinates of return points with a motion of a physical object, the motion being a combination of a translational motion and a rotational motion of a rigid body, and causing a driving path of the AV to be determined in view of the motion of the physical object.

Velocity estimation and object tracking for autonomous vehicle applications

Aspects of the disclosure relate to training and using a model for verifying accuracy of ICP alignments or alignments between data points using an iterative closest point algorithm. For instance, a model may be trained using ICP alignment data, including alignments between an object appearing in LIDAR sensor frames. The training may also include setting a definition for a trusted ICP alignment. In this regard, the model may be trained such that, n response to receiving additional LIDAR sensor frames and corresponding additional ICP alignment data, output a value indicative of whether the additional ICP alignment data is trusted according to the definition. The model may then be used to control a vehicle in an autonomous driving mode by determining whether alignment data for object determined using the ICP algorithm should be trusted.

Verification of iterative closest point alignments for autonomous vehicles

Example embodiments relate to calibration and localization of a light detection and ranging (lidar) device using a previously calibrated and localized lidar device. An example embodiment includes a method. The method includes receiving, by a computing device associated with a second vehicle, a first point cloud captured by a first lidar device of a first vehicle. The first point cloud includes points representing the second vehicle. The method also includes receiving, by the computing device, pose information indicative of a pose of the first vehicle. In addition, the method includes capturing, using a second lidar device of the second vehicle, a second point cloud. Further, the method includes receiving, by the computing device, a third point cloud representing the first vehicle. Yet further, the method includes calibrating and localizing, by the computing device, the second lidar device.

Calibration and localization of a light detection and ranging (lidar) device using a previously calibrated and localized lidar device

Aspects and implementations of the present disclosure address shortcomings of the existing technology by enabling efficient object identification and tracking in autonomous vehicle (AV) applications by using velocity data-assisted mapping of first set of points obtained for a first sensing data frame by a sensing system of the AV to a second set of points obtained for a second sensing data frame by the sensing system of the AV, the first set of points and the second set of points corresponding to an object in an environment of the AV, and causing a driving path of the AV to be determined in view of the performed mapping.

Doppler-assisted object mapping for autonomous vehicle applications

The technology relates to handling of self-reflections of sensor signals off of a portion of a vehicle that is operating in an autonomous driving mode. Vehicle pose information and sensor pose information are determined at a given point in time while operating in the autonomous driving mode. Return signals from one or more scans of the environment are received from onboard sensors such as lidar sensors. The system evaluates, based on the vehicle and sensor pose information, whether a segment between a given one of the one or more sensors and a received point from a selected one of the return signals crosses any surface of a 3D model of the vehicle. The received point is identified as a self-return point. In response to identifying the received point as a self-return point, the vehicle is able to perform a driving operation in the autonomous driving mode.

Self-reflection filtering

The technology relates to localizing a vehicle. As one approach, a first LIDAR sensor scan data of an environment of the vehicle and localization data for the vehicle at a location where the first LIDAR sensor scan was captured are stored. Thereafter, the computing device is suspended and subsequently unsuspended. After the computing device is unsuspended, second LIDAR sensor scan data of the vehicle's environment is received. The first LIDAR sensor scan data is compared to the second LIDAR sensor scan data to determine whether the vehicle has moved. Based on the determination of whether the vehicle has moved from the location, the stored localization data is used to localize the vehicle. Other approaches are also described.

Localization initialization for autonomous vehicles

The technology relates to enhancing or extending the field of view of sensors for vehicles configured to operate in an autonomous driving mode. One or more mirrors are used to reflect or redirect beams emitted from onboard sensors that would otherwise be wasted, for instance due to obstruction by a portion of the vehicle or because they are emitted at high pitch angles to the side. The mirrors are also used to redirect incoming beams from the external environment toward one or more of the onboard sensors. Using mirrors for such redirection can reduce or eliminate blind spots around the vehicle. A calibration system may be employed to account for mirror movement due to vibration or wind drag. Each mirror may be a front surface mirror. The mirrors may be positioned on the vehicle body, on a faring, or extending from a sensor housing on the vehicle.

Mirrors to extend sensor field of view in self-driving vehicles

Aspects and implementations of the present disclosure relate to filtering of return points from a point cloud based on radial velocity measurements. An example method includes: receiving, by a sensing system of an autonomous vehicle (AV), data representative of a point cloud comprising a plurality of return points, each return point comprising a radial velocity value and position coordinates representative of a reflecting region that reflects a transmission signal emitted by the sensing system; applying, to each of the plurality of return points, at least one threshold condition related to the radial velocity value of a given return point to identify a subset of return points within the plurality of return points; removing the subset of return points from the point cloud to generate a filtered point cloud; and identifying objects represented by the remaining return points in the filtered point cloud.

Filtering return points in a point cloud based on radial velocity measurement

A system includes a light detection and ranging device configured to generate, for each respective point of a plurality of points in an environment, a corresponding waveform that represents physical characteristics of the respective point. The system also includes a signal processor configured to determine, based on the corresponding waveform of each respective point, a map of the environment that includes a representation of a corresponding position of the respective point. The system additionally includes an embedding model configured to determine, for each respective point and based on the corresponding waveform, a corresponding vector comprising a plurality of values representative of the physical characteristics of the respective point. The system further includes a feature detector configured to detect or classify a physical feature based on (i) the corresponding positions of one or more points of the plurality of points and (ii) the corresponding vectors of the one or more points.

Laser waveform embedding

Methods and systems for laser point clouds are described herein. The method and system may include receiving, at a computing device, lidar data indicative of an environment of a vehicle from a first lidar data source, where the lidar data includes a first plurality of data points indicative of locations of reflections from the environment and further includes a respective intensity for each data point. The method and system also include determining a first surface normal for at least a first data point of the first plurality of data points. The method and system further includes determining a first angle of incidence for the first data point based on the surface normal. Additionally, the method and system includes adjusting the intensity of the first data point based on the first angle of incidence to create a first adjusted intensity for the first data point.

Growing community of inventors

Sunnyvale, CA, United States of America

Mingcheng Chen