Aeva, Inc.

Neato Robotics, Inc.

Apple Inc.

Bruno Hexsel

Kumar Bhargav Viswanatha

Jose Krause Perin

Krishna Toshniwal

Mina A Rezk

Steven Mathew Alexander

Heethesh Vhalve Naresh

Rajendra Tushar Moorti

James Nakamura

Heethesh Vhavle Naresh

Ahmad Al-Dahle

James Reuther

A set of POIs of a point cloud are received at a first filter. Each POI of the set of POIs is filtered. At a second filter, based on a first metric, a first score of the POI is determined. At a third filter, based on a second metric, a second score of the POI is determined. At the first filter, based on the first score and the second score, whether to accept the POI, modify the POI, or reject the POI, is determined to extract range or velocity information.

Techniques for parallel cascade filtering in point clouds

A set of POIs of a point cloud are received at a first filter, where each POI of the set of POIs comprises one or more points. Each POI of the set of POIs is filtered. A set of neighborhood points of a POI is selected. A metric for the set of neighborhood points is computed based on a property of the set of neighborhood points and the POI, wherein the property comprises a velocity. Based on the metric, whether to accept the POI, modify the POI, reject the POI, or transmit the POI to a second filter, to extract at least one of range or velocity information related to the target is determined. Provided the POI is not accepted, modified, or rejected, the POI is transmitted to the second filter to determine whether to accept, modify, or reject the POI to extract the at least one of range or velocity information related to the target.

Techniques for filtering points in a light detection and ranging system

A method generates first points based on a first scan of an environment that includes one or more moving objects. The method transforms the first points into a first static frame, which includes removing one or more of the first points corresponding to the one or more moving objects. The method generates second points based on a second scan of the environment that includes the one or more moving objects. The method transforms the second points into a second static frame, which includes removing one or more of the second points corresponding to the one or more moving objects. The method combines the first static frame and the second static frame into an accumulated static frame, which has an increase in resolution compared with the first static frame. The method then loads the accumulated static frame into a point cloud.

Techniques for point cloud frame accumulation in FMCW LIDAR

A method of operating a light detection and ranging (LiDAR) system is provided that includes performing a scene measurement using a LiDAR sensor capable of measuring Doppler per point. The method also includes estimating a velocity of the LiDAR sensor with respect to static points within the scene based on the scene measurement. The method may also include compensating for the velocity of the LiDAR sensor and compensating for a Doppler velocity of the LiDAR sensor.

Velocity estimation using doppler per point lidar systems

A system and method including, predicting a first set of Doppler velocity information for a first point set to produce a first predicted set of Doppler velocity information for the first point set; determining one or more Doppler velocity errors based on the first predicted set of Doppler velocity information and a second set of Doppler velocity information; predicting a third set of Doppler velocity information for the second point set to produce a second predicted set of Doppler velocity information for the second point set; determining one or more Doppler velocity errors based on at least one of the second predicted set of Doppler velocity information and a fourth set of Doppler velocity information; and producing a transform that aligns the one or more points of the first point set with the one or more points of the second point set.

Techniques for doppler point set registration

A system and method including, predicting, using a sensor system, a first set of Doppler velocity information for a first point set to produce a first predicted set of Doppler velocity information for the first point set. The method includes determining one or more Doppler velocity errors based on the first predicted set of Doppler velocity information and a second set of Doppler velocity information measured for one or more points of the first point set. The method includes producing, in view of the one or more Doppler velocity errors, a transform that aligns the one or more points of the first point set with one or more points of a second point set to produce one or more scan frames.

Velocity estimation using doppler per point LiDAR systems

Techniques for cascade filtering of a set of points of interest (POIs) in a light detection and ranging (LiDAR) system is described. The method includes performing a series of cascaded filtering of a set of points of interest (POIs) on a first point cloud. The method includes calculating, for each POI of the set of POIs, at least a first metric for a first set of neighborhood points to make a decision with respect to a subset of the set of POIs for transmission to a second point cloud. The method also includes extracting at least one of range or velocity information based on the second point cloud based on the decision.

Techniques for using a series cascaded filter to determine range and/or velocity information

A set of POIs of a point cloud are received at a first filter, where each POI of the set of POIs comprises one or more points. Each POI of the set of POIs is filtered. A set of neighborhood points of a POI is selected. A metric for the set of neighborhood points is computed. Based on the metric, whether to accept the POI, modify the POI, reject the POI, or transmit the POI to a second filter, to extract at least one of range or velocity information related to the target is determined. Provided the POI is accepted or modified, the POI is transmitted to a filtered point cloud; provided the POI is rejected, the POI is prevented from reaching the filtered point cloud; provided the POI is not accepted, modified, or rejected, the POI is transmitted to a second filter.

Techniques for point cloud filtering

Techniques for cascade filtering of a set of points of interest (POIs) in a light detection and ranging (LiDAR) system is described. The method includes performing a series of cascaded filtering of a set of points of interest (POIs) in a first point cloud and a second point cloud of the LiDAR system. The method also includes calculating at least a first metric over at least the POI and to make a decision with respect to the second point cloud, and extracting at least one of range and velocity information based on the second point cloud.

Techniques for cascade filtering of points of interest in point clouds

Some embodiments provide an autonomous navigation system which autonomously navigates a vehicle through an environment based on predicted trajectories of one or more separate dynamic elements through the environment. The system identifies contextual cues associated with a monitored dynamic element, based on features of the dynamic element and actions of the dynamic element relative to various elements of the environment, including motions relative to other dynamic elements. A monitored dynamic element can be associated with a particular intention, which specifies a prediction of dynamic element movement through the environment, based on a correlation between identified contextual cues associated with the monitored dynamic element and a set of contextual cues which are associated with the particular intention. A predicted trajectory of the dynamic element is generated based on an associated intention. A targeted signal can be directed to a target dynamic element based on a predicted trajectory of the dynamic element.

Intention recognition

A set of POIs of a point cloud are received at a first filter, where each POI of the set of POIs comprises one or more points. Each POI of the set of POIs is filtered. At a second filter, a first set of neighborhood points of a POI is selected. A first metric for the first set of neighborhood points is computed. Based on the first metric, a first score of the POI is determined. At a third filter, a second set of neighborhood points of a POI is selected. A second metric for the second set of neighborhood points is computed. Based on the second metric, a second score of the POI is determined. At the first filter, based on the first score and the second score, whether to accept the POI, modify the POI, or reject the POI, is determined to extract range or velocity information.

Techniques for fast point cloud filtering using a parallel cascaded filter

A system and method including, determining an alignment between one or more points of a first point set and one or more points of a second point set; predicting a first set of Doppler velocity information for the first point set to produce a first predicted set of Doppler velocity information; comparing a second set of Doppler velocity information and the first predicted set of Doppler velocity information to compute one or more Doppler velocity errors; predicting a third set of Doppler velocity information for the second point set to produce a second predicted set of Doppler velocity information; comparing a fourth set of Doppler velocity information and the second predicted set of Doppler velocity information to compute Doppler velocity errors; and producing a transform that aligns the one or more points of the first point set with the one or more points of the second point set.

A system and method including, determining an alignment between one or more points of a first point set and one or more points of a second point set; provided the alignment does not satisfy a point set alignment threshold, wherein the point set alignment threshold represents a geometric alignment error threshold between the first and second point sets: predicting a first set of Doppler velocity information for the first point set to produce a first predicted set of Doppler velocity information for the first point set; comparing a second set of Doppler velocity information and the first predicted set of Doppler velocity information to compute one or more Doppler velocity errors; and producing a transform that aligns the one or more points of the first point set with the one or more points of the second point set to produce a globally consistent registered point set.

Each POI of a set of POIs of a first point cloud is filtered. At a first filter, a first set of neighborhood points of a POI is selected; a first metric for the first set of neighborhood points is computed; and based on the first metric, whether to accept, modify, reject or transmit the POI to a second filter is determined. Provided the POI is accepted or modified, the POI is transmitted to a second point cloud; provided the POI is rejected, the POI is prevented from reaching the second point cloud; provided the POI is not accepted, modified, or rejected, the POI is transmitted to a second filter. At the second filter, provided the POI is accepted or modified, the POI is transmitted to the second point cloud. At least one of range and velocity information is extracted based on the second point cloud.

Growing community of inventors

Mountain View, CA, United States of America

Bruno Hexsel