Canoo Technologies Inc.

University of Southern California

Jongmoo Choi

Kilsoo Kim

David R Arft

Aviral K Singh

Mayukh Sattiraju

Siddharth Agarwal

Gerard Medioni

Dheemanth Uppalapati

Paolo E Pucci

Dae Jin Kim

Ruizhe Wang

Lung En Jan

Emilio Aron Moyers Barrera

Vishal U Shanbhag

Lei Cao

Shanmukha M Bhumireddy

Phillip Vu

Emilio Moyers

Rishitha Kalakata

Dhruvil Darji

A method includes obtaining sensor data associated with a target object at a host vehicle and identifying relative positions of the target object with reference to the host vehicle over time. The method also includes generating polynomials associated with the target object over time based on the relative positions, where coefficients of each polynomial are based on a magnitude of a covariance of the relative positions used to generate the polynomial. The method further includes generating a weighted combination of the polynomials for the target object, where the weighted combination is a representation of an estimated behavior of the target object. In addition, the method includes determining whether a collision between the host vehicle and the target object is possible based on the weighted combination of the polynomials and, in response to determining that the collision is possible, initiating one or more corrective actions by the host vehicle.

System and method for target behavior prediction in advanced driving assist system (ADAS), autonomous driving (AD), or other applications

A method includes obtaining, using at least one processing device, a vanishing point and a boundary based on image data associated with a scene, where the boundary is associated with a detected object within the scene. The method also includes repeatedly, during multiple iterations and using the at least one processing device, (i) identifying multiple patches within the boundary and (ii) determining a similarity of the image data contained within the multiple patches. The method further includes identifying, using the at least one processing device, a modification to be applied to the boundary based on the identified patches and the determined similarities. In addition, the method includes generating, using the at least one processing device, a refined boundary based on the modification, where the refined boundary identifies a specified portion of the detected object.

Vanishing point determination, symmetry-based boundary refinement, and component detection for vehicle object detection or other applications

A method includes obtaining, using at least one processing device, image data associated with a scene. The method also includes identifying, using the at least one processing device, multiple line segments based on the image data. The method further includes identifying, using the at least one processing device, one or more boundaries around one or more objects detected in the image data. In addition, the method includes estimating, using the at least one processing device, a position of a vanishing point associated with the image data based on multiple collections of the line segments while excluding, from the multiple collections, one or more of the line segments that overlap with or that are included within the one or more boundaries.

An apparatus includes an inertial measurement unit (IMU) configured to detect motion characteristics of a vehicle. The apparatus also includes an odometry system configured to detect a wheel speed of each wheel of the vehicle. The apparatus further includes at least one processor communicatively connected to the IMU and the odometry system, the at least one processor configured to determine first parameters for predicting a path of the vehicle, determine second parameters for predicting the path of the vehicle, and predict the path of the vehicle using a combination of the first parameters and the second parameters, wherein the combination is weighted based on a longitudinal acceleration of the vehicle obtained using the IMU.

System and method in vehicle path prediction based on odometry and inertial measurement unit

A method includes identifying, using at least one processor, a first point associated with an uncertain location of an object in a space and a polynomial curve associated with an uncertain location of a feature in the space. The method also includes determining, using the at least one processor, a probabilistic proximity of the object and the feature. The probabilistic proximity is determined by identifying a second point on the polynomial curve, transforming an uncertainty associated with the polynomial curve into an uncertainty associated with the second point, and identifying the probabilistic proximity of the object and the feature using the first and second points and the uncertainty associated with the second point.

Proximity detection for automotive vehicles and other systems based on probabilistic computing techniques

An apparatus includes at least one camera configured to capture an image of a traffic lane in front of a vehicle. The apparatus also includes a radar transceiver configured to detect one or more target vehicles proximate to the vehicle. The apparatus further includes a path prediction and vehicle detection controller configured to determine first parameters for predicting a path of the vehicle; determine second parameters for predicting the path of the vehicle; predict the path of the vehicle using a combination of the first parameters and the second parameters, where the combination is weighted based on a speed of the vehicle; identify one of the one or more target vehicles as a closest in path vehicle based on the predicted path of the vehicle; and activate at least one of a braking control and a steering control based on a proximity of the identified closest in path vehicle.

System and method in vehicle path prediction based on full nonlinear kinematics

An apparatus includes at least one camera configured to capture a series of image frames for traffic lanes in front of an ego vehicle, where each of the series of image frames is captured at a different one of a plurality of times. A target object detection and tracking controller is configured to process each of the image frames using pixel measurements extracted from the respective image frame to determine, from the pixel measurements, a predicted time to line crossing for a target vehicle detected in the respective image frame at a time corresponding to capture of the respective image frame.

System and method in the prediction of target vehicle behavior based on image frame and normalization

A method includes obtaining a fisheye image of a scene and identifying multiple regions of interest in the fisheye image. The method also includes applying one or more transformations to transform and rotate one or more of the regions of interest in the fisheye image to produce one or more transformed regions. The method further includes generating a collage image having at least one portion based on the fisheye image and one or more portions containing the one or more transformed regions. In addition, the method includes performing object detection to identify one or more objects captured in the collage image.

Fisheye collage transformation for road object detection or other object detection

An apparatus includes at least one camera configured to capture an image of a traffic lane in front of a vehicle. The apparatus also includes a vehicle behavior prediction controller configured to determine lane boundaries and road curvature for a segment of a traffic lane occupied by the vehicle from the captured image and prior captured images; determine lateral distances of the vehicle from the lane boundaries and a rate of departure of the vehicle from the occupied traffic lane that is accurate for the determined road curvature; determine a time to line crossing for the vehicle from the lateral distances and the rate of departure; and activate a lane departure warning indicator based on the determined time to line crossing.

System and method in lane departure warning with full nonlinear kinematics and curvature

A method includes identifying a path to be followed by an ego vehicle. The method also includes determining a desired yaw rate and a desired yaw acceleration for the ego vehicle based on the identified path. The method further includes determining a desired yaw moment for the ego vehicle based on the desired yaw rate and the desired yaw acceleration. In addition, the method includes distributing the desired yaw moment to multiple wheels of the ego vehicle such that the distributed desired yaw moment creates lateral movement of the ego vehicle during travel along the identified path. In some cases, the desired yaw rate and the desired yaw acceleration for the ego vehicle may be determined based on nonlinear kinematics of the ego vehicle, and the desired yaw moment for the ego vehicle may be determined based on a single-track dynamic model of the ego vehicle.

Autonomous lateral control of vehicle using direct yaw moment control

A method includes determining a desired yaw moment to be applied to an ego vehicle during travel. The method also includes identifying yaw moment changes that are achievable using different torque vectoring techniques supported by the ego vehicle. The method further includes selecting at least one of the torque vectoring techniques based on the identified yaw moment changes. In addition, the method includes using the at least one selected torque vectoring technique to obtain the desired yaw moment and create lateral movement of the ego vehicle during the travel. In some cases, a desired response time associated with the lateral movement of the ego vehicle may be used, where steering control provides a faster response time and torque vectoring control provides a slower response time. The at least one torque vectoring technique may be selected based on different energy efficiencies associated with different ones of the torque vectoring techniques.

Path tracking control for self-driving of vehicle with yaw moment distribution

A method includes obtaining sensor data associated with a target object at an ego vehicle, where the sensor data includes image data within an image frame. The method also includes identifying a first longitudinal distance of a first point of the target object from the ego vehicle within a world frame. The method further includes identifying a first lateral distance of the first point of the target object from the ego vehicle within the world frame using a first normalization ratio that is based on the image data. The method also includes predicting a future path of the target object based on the first longitudinal distance and the first lateral distance. In addition, the method includes controlling at least one operation of the ego vehicle based on the predicted future path of the target object.

Prediction of target object's behavior based on world and image frames

An apparatus includes at least one camera configured to capture at least one image of a traffic lane, an inertial measurement unit (IMU) configured to detect motion characteristics, and at least one processor. The at least one processor is configured to obtain a vehicle motion trajectory using the IMU and based on one or more vehicle path prediction parameters, obtain a vehicle vision trajectory based on the at least one image, wherein the vehicle vision trajectory includes at least one lane boundary, determine distances between one or more points on the vehicle and one or more intersection points of the at least one lane boundary based on the obtained vehicle motion trajectory, determine at least one time to line crossing (TTLC) based on the determined distances and a speed of the vehicle, and activate a lane departure warning indicator based on the determined at least one TTLC.

System and method for lane departure warning with ego motion and vision

The present disclosure describes systems and techniques relating to generating three dimensional (3D) models from range sensor data. According to an aspect, frames of range scan data captured using one or more three dimensional (3D) sensors are obtained, where the frames correspond to different views of an object or scene; point clouds for the frames are registered with each other by maximizing coherence of projected occluding boundaries of the object or scene within the frames using an optimization algorithm with a cost function that computes pairwise or global contour correspondences; and the registered point clouds are provided for use in 3D modeling of the object or scene. Further, the cost function, which maximizing contour coherence, can be used with more than two point clouds for more than two frames at a time in a global optimization framework.

Three-dimensional modeling from wide baseline range scans

The present disclosure describes systems and techniques relating to generating three dimensional (3D) models from range sensor data. According to an aspect, multiple 3D point clouds, which are captured using one or more 3D cameras, are obtained. At least two of the 3D point clouds correspond to different positions of a body relative to at least a single one of the one or more 3D cameras. Two or more of the 3D point clouds are identified as corresponding to two or more predefined poses, and a segmented representation of the body is generated, in accordance with a 3D part-based volumetric model including cylindrical representations, based on the two 3D point clouds identified as corresponding to the two predefined pose.

Growing community of inventors

Gardena, CA, United States of America

Jongmoo Choi