Abstract: Techniques are described for generating bird's eye view (BEV) images and segmentation maps. According to one or more embodiments, a system is provided comprising a processor that executes computer executable components stored in at least one memory, comprising a machine learning component that generates a synthesized bird's eye view image from a stitched image based on removing artifacts from the stitched image present from a transformation process. The system further comprising a generator that produces the synthesized bird's eye view image and a segmentation map, and a discriminator that predicts whether the synthesized bird's eye view image and the segmentation map are real or generated.

Abstract: Systems, devices, computer-implemented methods, and/or computer program products that can facilitate pedestrian detection via a boundary cylinder model are addressed. In one example, a system can comprise a processor that executes computer executable components stored in memory. The computer-executable components can comprise a bounding cylinder model component that determines numerical values for height and radius parameters of a three-dimensional bounding cylinder model representing a pedestrian, and that generates the three-dimensional bounding cylinder model representing the pedestrian based on the numerical values.

Abstract: Various systems and methods are presented regarding utilizing technology onboard a vehicle to identify and annotate a vehicle parking spot. The parking spot is identified in a digital image captured by an onboard camera and imaging system, wherein an image including the parking spot can be annotated with a shape identifying the parking spot and further annotated with information regarding the parking spot, e.g., dimensions of the parking spot, an entrance to the parking spot, whether the parking spot is occupied, a location of the parking spot, and suchlike. The parking spot information can be transmitted to a remotely located vehicle and/or a remotely located computer system configured to compile information regarding the parking spot.

Abstract: A cross-range 3D object detection method and system operable for training a 3D object detection model with N sub-groups of a point cloud corresponding to N detection distance ranges to form N 3D object detection models forming an ensemble 3D object detection model. Training the 3D object detection model with the N sub-groups of the point cloud corresponding to the N detection distance ranges includes training the 3D object detection model progressively from distant to near. Training the 3D object detection model with the N sub-groups of the point cloud corresponding to the N detection distance ranges includes, each time the 3D object detection model converges, saving resulting weights and adding a corresponding network to the ensemble 3D object detection model.

Abstract: One or more embodiments herein can provide a process to determine dead-reckoning localization of a vehicle. An exemplary system can comprise a memory that stores computer executable components, and a processor that executes the computer executable components stored in the memory, wherein the computer executable components can comprise an obtaining component that obtains plural sensor readings defining movement of a vehicle, wherein the plural sensor readings comprise an inertial sensor reading, a kinematics sensor reading, and an odometry sensor reading, and a generation component that generates, based on the plural sensor readings, a pose value defining a position of the vehicle relative to an environment in which the vehicle is disposed. A sensing sub-system of the exemplary system can comprise an inertial measurement unit sensor, a kinematics sensor, and an odometry sensor.

Abstract: One or more embodiments herein can enable identification of an obstacle free area about an object. An exemplary system can comprise a memory that stores computer executable components, and a processor that executes the computer executable components stored in the memory, wherein the computer executable components can comprise an obtaining component that obtains raw data defining a physical state of an environment around an object from a vantage of the object, and a generation component that, based on the raw data, generates a dimension of a portion or more of a virtual polygon representing a boundary about the object, wherein the boundary bounds free space about the object. A sensing sub-system can comprise both an ultrasonic sensor and a camera that can separately sense the environment about the object from the vantage of the object to thereby generate separate polygon measurement sets.

Abstract: A system and method, including: a data input device coupled to a vehicle, wherein the data input device is operable for gathering input data related to one or more of an environmental context surrounding or within the vehicle, an occupant state within the vehicle, a vehicle state, a location of the vehicle, and a noise condition inside or outside of the vehicle; an artificial intelligence system coupled to the data input device, wherein the artificial intelligence system is operable for generating a sound qualifier based on the gathered input data related to the one or more of the environmental context surrounding or within the vehicle, the occupant state within the vehicle, the vehicle state, and the noise condition inside or outside of the vehicle; and an audio generation module operable for receiving the sound qualifier from the artificial intelligence system and synthesizing a soundscape based on the sound qualifier.

Abstract: Various systems and methods are presented regarding utilizing technology onboard a vehicle to identify and annotate a vehicle parking spot. The parking spot is identified in a digital image captured by an onboard camera and imaging system, wherein an image including the parking spot can be annotated with a shape identifying the parking spot and further annotated with information regarding the parking spot, e.g., dimensions of the parking spot, an entrance to the parking spot, whether the parking spot is occupied, a location of the parking spot, and suchlike. The parking spot information can be transmitted to a remotely located vehicle and/or a remotely located computer system configured to compile information regarding the parking spot.

Abstract: Methods and systems for unsupervised depth estimation for fisheye cameras using spatial-temporal (and, optionally, modal) consistency. This unsupervised depth estimation works directly on raw, distorted stereo fisheye images, such as those obtained from the four fisheye camera disposed around a vehicle in rigid alignment. Temporal consistency involves training a depth estimation model using a sequence of frames as input, while spatial consistency involves training the depth estimation model using overlapping images from synchronized stereo camera pairs. Images from different stereo camera pairs can also be used at different times. Modal consistency, when applied, dictates that different sensor types (e.g., camera, lidar, etc.) must also agree. The methods and systems of the present disclosure utilize a fisheye camera projection model that projects a disparity map into a point cloud map, which aides in the rectification of stereo pairs.

Abstract: A method and system by which a bounding box disposed around a segmented object in a camera (or other perception sensor) 2D image can be used to produce an estimate for both the location of the object—its position relative to the position of the camera that obtained the image (i.e., translation)—and the angle of rotation of the surface that the object is located on. The method and system may be used by an advanced driver assistance system (ADAS), an autonomous driving (AD) system, or the like. The input includes a simple camera (or other perception sensor) 2D image, with the ego vehicle generating 2D or 3D bounding boxes for objects detected at the scene. The output includes, for each object, its estimated distance from the ego vehicle camera/perception sensor and the angle of rotation of the surface underneath the object relative to the surface underneath the ego vehicle.

Abstract: A LIDAR-based method of determining an absolute speed of an object at a relatively longer distance from an ego vehicle, including: estimating a self speed of the ego vehicle using a first frame t?1 and a second frame t obtained from a LIDAR sensor by estimating an intervening rotation ? about a z axis and translation in orthogonal x and y directions using a deep learning algorithm over a relatively closer distance range; dividing each of the first frame t?1 and the second frame t into multiple adjacent input ranges and estimating a relative speed of the object at the relatively longer distance by subsequently processing each frame using a network, with each input range processed using a corresponding convolutional neural network; and combining the estimation of the estimating the self speed with the estimation of the estimating the relative speed to obtain an estimation of the absolute speed.

Abstract: A system and method, including: a data input device coupled to a vehicle, wherein the data input device is operable for gathering input data related to one or more of an environmental context surrounding or within the vehicle, an occupant state within the vehicle, a vehicle state, a location of the vehicle, and a noise condition inside or outside of the vehicle; an artificial intelligence system coupled to the data input device, wherein the artificial intelligence system is operable for generating a sound qualifier based on the gathered input data related to the one or more of the environmental context surrounding or within the vehicle, the occupant state within the vehicle, the vehicle state, and the noise condition inside or outside of the vehicle; and an audio generation module operable for receiving the sound qualifier from the artificial intelligence system and synthesizing a soundscape based on the sound qualifier.

Abstract: One or more embodiments herein can provide a process to determine dead-reckoning localization of a vehicle. An exemplary system can comprise a memory that stores computer executable components, and a processor that executes the computer executable components stored in the memory, wherein the computer executable components can comprise an obtaining component that obtains plural sensor readings defining movement of a vehicle, wherein the plural sensor readings comprise an inertial sensor reading, a kinematics sensor reading, and an odometry sensor reading, and a generation component that generates, based on the plural sensor readings, a pose value defining a position of the vehicle relative to an environment in which the vehicle is disposed. A sensing sub-system of the exemplary system can comprise an inertial measurement unit sensor, a kinematics sensor, and an odometry sensor.

Abstract: One or more embodiments herein can enable identification of an obstacle free area about an object. An exemplary system can comprise a memory that stores computer executable components, and a processor that executes the computer executable components stored in the memory, wherein the computer executable components can comprise an obtaining component that obtains raw data defining a physical state of an environment around an object from a vantage of the object, and a generation component that, based on the raw data, generates a dimension of a portion or more of a virtual polygon representing a boundary about the object, wherein the boundary bounds free space about the object. A sensing sub-system can comprise both an ultrasonic sensor and a camera that can separately sense the environment about the object from the vantage of the object to thereby generate separate polygon measurement sets.

Abstract: A method, software tool, and system for generating realistic synthetic ride-requests associated with a mobility or transportation service, including: utilizing a generative adversarial network, learning the spatial-temporal distribution of a plurality of real ride-requests; and, utilizing the generative adversarial network and based on the learning step, generating one or more synthetic source and destination ride-request geolocations that retain a statistical distribution of the plurality of real ride-requests. The generative adversarial network is a Wasserstein generative adversarial network.

Abstract: Techniques are described for generating bird's eye view (BEV) images and segmentation maps. According to one or more embodiments, a system is provided comprising a processor that executes computer executable components stored in at least one memory, comprising a machine learning component that generates a synthesized bird's eye view image from a stitched image based on removing artifacts from the stitched image present from a transformation process. The system further comprising a generator that produces the synthesized bird's eye view image and a segmentation map, and a discriminator that predicts whether the synthesized bird's eye view image and the segmentation map are real or generated.

Abstract: Systems and methods for fisheye camera calibration and BEV image generation in a simulation environment. This fisheye camera calibration enables the extrinsic and intrinsic parameters of the fisheye camera to be computed in the simulation environment, where data is readily available, collectible, and manipulatable. Given a surround vision system, with multiple fisheye cameras disposed around a vehicle, and these extrinsic and intrinsic parameters, undistorted and BEV images of the surroundings of the vehicle can be generated in the simulated environment, for simulated fisheye camera testing and validation, which may then be extrapolated to real-world fisheye camera testing and validation, as appropriate. Because the simulation tool can be used to create and readily manipulate the simulated fisheye camera, the vehicle, its surroundings, obstacles, targets, markers, and the like, the entire calibration and image generation process is streamlined and may be automated.

Abstract: A stereo pair camera system for depth estimation, including: a first camera disposed in a first position along a longitudinal axis, a lateral axis, and a vertical axis and having a first field of view; and a second camera disposed in a second position along the longitudinal axis, the lateral axis, and the vertical axis and having a second field of view; wherein the first camera is a of a first type and the second camera is of a second type that is different from the first type; and wherein the first field of view overlaps with the second field of view. Optionally, the first position is spaced apart from the second position along one or more of the longitudinal axis and the vertical axis. The depth estimation is used by one or more of a driver assist system and an autonomous driving system of a vehicle.

Abstract: The present invention relates to a driver assistance (DA) system and method for vehicle flank safety. This system and method also have utility in autonomous driving (AD) applications. The system and method optionally pair one or more cameras with one or more proximity sensors to provide free space awareness and contact avoidance related to the flank of a vehicle, providing a virtual distance grid overlay on one or more camera views available to an operator, predefined safety boundary intrusion detection under low-speed maneuvering conditions, door-open obstruction and danger detection, and a flank illumination system.

Abstract: Systems, devices, computer-implemented methods, and/or computer program products that can facilitate free space detection using fast sensor fusion are provided. In one example, a system can comprise a processor that executes computer executable components stored in memory. The computer executable components can comprise a threshold component, a pixel fusion component, and a frame control component. The threshold component can update the count of a plurality of pixel counters based on confidence scores of at least one camera and at least one sensor in a single pixel. The pixel fusion component can perform sensor fusion on the confidence scores based on a determination that the count of the plurality of pixel counters is less than a defined threshold. The frame control component can bypass the sensor fusion on the confidence scores based on a determination that the count of the plurality of pixel counters is greater than a defined threshold.