Abstract: A hand trajectory recognition method for a following robot based on hand velocity and trajectory distribution comprises: sampling and photographing an operator by a kinect camera to obtain hand projection plane data; smoothing the hand projection plane data by moving average, establishing velocity vectors, and processing the velocity vectors to obtain a hand movement descriptor; establishing a hand movement area, traversing hand three-dimensional positions of all frames in an order of sampling and photographing, assigning a mesh where the hand three-dimensional position of each frame is located, and calculating centroid positions of all assigned meshes; establishing centroid directing vectors, and processing the centroid directing vectors to obtain a hand trajectory shape descriptor; and processing cosine values of two angles to obtain a common similarity of the movement descriptor and the trajectory shape descriptor to standard descriptors, and using a standard gesture with a maximum common similarity as a r

Abstract: When a main power from a main battery of a vehicle is interrupted (or cut off), an auxiliary power supplied from an auxiliary battery may be used to drive electrical components of the vehicle. The vehicle includes: a main battery to supply the main power; the auxiliary battery to supply an auxiliary power; and an image recording controller to operate with the main power, capture and record an image of surroundings of the vehicle. In particular, when the supply of the main power is interrupted, the image recording controller performs capturing and recording the image by receiving the auxiliary power, controls a supply path of the auxiliary power so that the auxiliary power is supplied to a predetermined electrical component of the vehicle to continuously operate the predetermined electrical component using the auxiliary power.

Abstract: A method and an apparatus for detecting a lane is provided. The lane detection apparatus according to an embodiment includes: an acquisition unit configured to acquire a front image of a vehicle; and a processor configured to input the image acquired through the acquisition unit to an AI model, and to detect information of a lane on a road, and the AI model is trained to detect lane information that is expressed in a plane form from an input image. Accordingly, data imbalance between a lane area and a non-lane area can be solved by using the AI model which learns/predicts lane information that is expressed in a plane form, not in a segment form such as a straight line or curved line.

Abstract: A driving support system for a vehicle includes an information obtainer, a target vehicle speed calculator, a comparison processor, and a determination processor. The information obtainer is configured to obtain curvature information about a curvature of a driving lane of the vehicle and about a curvature of an adjacent lane. The adjacent lane is adjacent to the driving lane. The target vehicle speed calculator is configured to calculate a target vehicle speed in the driving lane and a target vehicle speed in the adjacent lane using the curvature information. The comparison processor is configured to compare a difference between the target vehicle speed in the driving lane and the target vehicle speed in the adjacent lane with a threshold. If the difference is greater than or equal to the threshold, the determination processor is configured to determine that a road branches off in a traveling direction of the vehicle.

Abstract: A method for calibrating at least one position sensor in a vehicle, comprising: saving a map in the vehicle with at least one item of virtual location information from the environment of a predefined calibration route, moving the vehicle on the predefined calibration route, detecting at least one item of actual location information via the at least one position sensor while the vehicle moves on the predefined calibration route, carrying out a comparison between the at least one item of virtual location information and the at least one item of actual location information, calculating a map-based movement of the vehicle within the saved map based on the comparison, calculating an odometry-based movement of the vehicle, and calibrating the at least one position sensor based on a comparison between the map-based movement and the odometry-based movement.

Abstract: Example implementations are provided for an arrangement of co-aligned rotating sensors. One example device includes a light detection and ranging (LIDAR) transmitter that emits light pulses toward a scene according to a pointing direction of the device. The device also includes a LIDAR receiver that detects reflections of the emitted light pulses reflecting from the scene. The device also includes an image sensor that captures an image of the scene based on at least external light originating from one or more external light sources. The device also includes a platform that supports the LIDAR transmitter, the LIDAR receiver, and the image sensor in a particular relative arrangement. The device also includes an actuator that rotates the platform about an axis to adjust the pointing direction of the device.

Abstract: A system, method, and computer-program product includes detecting, via a localization machine learning model, a target object within a scene based on downsampled image data of the scene, identifying a likely position of the target object within original image data of the scene, extracting, from the original image data of the scene, a target sub-image containing the target object, classifying, via an object classification machine learning model, the target object to a probable object class of a plurality of distinct object classes, routing the target image resolution of the target sub-image to a target object-condition machine learning classification model of a plurality of distinct object-condition machine learning classification models, classifying, via the target object-condition machine learning classification model, the target object to a probable object-condition class, and displaying, via a graphical user interface, a representation of the target object in association with the probable object-condition

Abstract: A method includes calculating a stereo-disparity map between the initial image and the final image, for each column of the stereo-disparity map, calculating an average value of the stereo-disparities of the pixels of the column, calculating the slope and/or constant factor of a linear function approximating variations of said average values; and calculating said information relative to the relative speed between the object and the camera, based on the slope and/or the constant factor.

Abstract: A system for asset identification and mapping is configured to information related to a plurality of objects with at least one sensor. The captured information may be processed to identify one or more assets among the objects, and a map generated including a geographic area proximate to at least one of the one or more assets. The at least one of the one or more assets may then be identified on the map.

Abstract: Sensor data and data from V2V messages are used to detect obstacles. Additional obstacles are identified in map data according to a vehicle's position. In response to detecting a turn intent, a controller calculates a turn path according to the detected obstacles and properties of the vehicle. The turn path is presented to a user, such as by display on a HUD or superimposing the turn path on an image from a forward facing camera. A desired steering angle to traverse the turn path may be displayed, such as relative to the current steering angle of the vehicle. Notifications regarding the path and turning intent of other vehicle may be displayed along with the turn path.

Abstract: A light inspection system positions an autonomous vehicle (AV) in a field of view of a camera such that the camera captures an image of a light of the AV. The light inspection system instructs a camera to capture an image of the light while the light is switched on. The light inspection system receives the captured image, determines a luminance of the light based on the image, and determines to service the light in response to the luminance of the light being below a threshold luminance.

Abstract: A system for capturing location-based data includes an interface and a processor. The interface is configured to receive a location-based data description for capturing the location-based data. The processor is configured to determine a location based at least in part on the location-based data description, create a location-based data identification job based at least in part on the location, cause execution of the location-based data identification job to a set of vehicle event recorder systems, wherein the location-based data identification job is executed by a vehicle event recorder system of the set of vehicle event recorder systems to acquire sensor data in response to a vehicle being able to acquire the sensor data related to the location of the location-based data identification job, receive the sensor data from the vehicle event recorder system, and determine the location-based data based at least in part on the sensor data.

Abstract: Systems and methods described herein relate to training a machine-learning-based monocular depth estimator.

Abstract: A lane keeping assist system includes: a camera configured to provide an image around a vehicle as image information; a lane information generator configured to generate image reliability information and first lane information, based on the image information; an image storage configured to store the image information for each predetermined time among predetermined times; a neural network learning device configured to generate second lane information based on the image reliability information and the image information stored for each predetermined time; and a steering controller configured to select either one of the first lane information and the second lane information as lane information, based on the image reliability information, and generate steering information based on the selected lane information.

Abstract: System and techniques for vehicle environment modeling with a camera are described herein. A time-ordered sequence of images representative of a road surface may be obtained. An image from this sequence is a current image. A data set may then be provided to an artificial neural network (ANN) to produce a three-dimensional structure of a scene. Here, the data set includes a portion of the sequence of images that includes the current image, motion of the sensor from which the images were obtained, and an epipole. The road surface is then modeled using the three-dimensional structure of the scene.

Abstract: A map update system includes a data transmission device mounted on a vehicle and a map server which stores map data. The data transmission device generates sensor data representing a road environment of surroundings of the vehicle in a predetermined position, calculates a matching degree between the road environment of the surroundings of the vehicle and a road environment in the predetermined position represented by the map data, and causes a communication circuit mounted on the vehicle to transmit the sensor data and information representing the matching degree to the map server. The map server transmits the map data by utilizing sensor data, among the sensor data received via a communication device, having a matching degree less than a matching degree threshold with a higher priority than sensor data having a matching degree or greater than the matching degree threshold.

Abstract: A method for measuring an inter-vehicle distance using a processor is provided. The method includes acquiring a driving image photographed by a photographing device of a first vehicle; detecting a second vehicle from the driving image and calculating a ratio between an image width of the detected second vehicle and an image width of a lane in which the second vehicle is located; determining a size class of the second vehicle among a plurality of size classes based on the calculated ratio; determining a width of the second vehicle based on the determined size class of the second vehicle; and calculating a distance from the photographing device to the second vehicle based on the determined width of the second vehicle, a focal length of the photographing device, and the image width of the second vehicle.

Abstract: In a traffic emergency, there is no time for a human to integrate multiple sensor data streams and devise a plan for avoiding a collision. Only the electronic reflexes of a trained automatic system can provide evasive action in time. Disclosed is an artificial intelligence (AI) model trained to recognize an imminent collision based on sensor data, rapidly devise and test a large number of possible sequences of actions, some drawn from a library of previously-successful strategies and others invented by the AI model. If any sequence can avoid the collision, the AI model implements that sequence immediately. If none of the sequences can avoid the collision, the AI model calculates the harm caused by each sequence and picks the one that causes the least harm (fatalities, injuries, etc.) for implementation. AI is needed to find a possible solution in time to implement it and thereby mitigate the imminent collision.

Abstract: A computer-implemented method for checking the correct deployment of an airbag device comprises, in a vehicle or test vehicle model, optionally arranging a dummy or model of a vehicle occupant in the vicinity of the airbag device, triggering the deployment of the air chamber of the airbag device, taking at least one image of a scene including the air chamber at a plurality of discrete times following the triggering of its deployment and verifying the volume of inflation and/or deployment of the air chamber at the plurality of discrete times following the triggering of its deployment on the basis of the at least one image.

Abstract: A list for accurately identifying objects with different sizes that are the same attributes can be generated automatically. A classification unit classifies, from a group of images consisting of images including objects with any of a plurality of attributes, each of the images including the objects that have an identical attribute and have different real sizes into an identical cluster. An output unit outputs each of clusters into which at least two of the images are classified as a list of images with different real sizes of the objects for the identical attribute.

Abstract: A measurement apparatus operable to perform measurement concerning a predetermined item for a target subject which is a measurement target. The apparatus comprises an acquisition unit configured to, for a captured image acquired by capturing of an image capturing range that includes the target subject, acquire distance information indicating a distribution of a subject distance from an image capturing apparatus that performed the capturing and, at least, normal line information that indicates a normal line of a surface of the target subject; and a presentation unit configured to present a notification that indicates an image capturing direction of the image capturing apparatus for performing measurement of the predetermined item based on the normal line information acquired by the acquisition unit.

Abstract: Measuring speed of a vehicle in a road environment. During calibration, multiple images are captured of a calibration vehicle traveling at a known ground speed. A calibration image feature is located in the image of the calibration vehicle. An optical flow of the calibration image feature is computed to determine a model between an image speed of the calibration image feature and the known ground speed of the calibration vehicle. During speed measurement, multiple images are captured of a target vehicle traveling along a road surface at unknown ground speed. A target image feature may be located in an image of the target vehicle. An image speed may be computed of the target image feature. The model may be applied to determine the ground speed of the target vehicle from the image speed of the target image feature.

Abstract: A system, such as a driver assistance system, arranged to form images of outside a vehicle for viewing from an eye box region within the vehicle. The system comprises an image capture device, a picture generating unit and an optical element. The image capture device is arranged to be mounted to the vehicle. The image capture device is arranged to capture images outside the vehicle. The optical element is disposed in front of the replay plane. The optical element has a focal length. The distance between the replay plane and the optical element is less that the focal length of the optical element such that the optical element forms a virtual image of each picture for viewing from the eye box region. The picture generating unit may be a holographic projector. The holographic projector may be arranged to form pictures on a replay plane. Each picture may correspond to an image captured by the image capture device.

Abstract: A method and apparatus for recognizing an object are provided, including extracting a feature from an input image and generating a feature map in a neural network. In parallel with the generating of the feature map, a region of interest (ROI) corresponding to an object of interest is extracted from the input image, and a number of object candidate regions used to detect the object of interest is determined based on a size of the ROI. The object of interest is recognized from the ROI based on the number of object candidate regions in the neural network.

Abstract: A system for mapping traffic lights and for determining traffic light relevancy for use in autonomous vehicle navigation. The system may include at least one processor programmed to: receive at least one location identifier associated with a traffic light; receive a state identifier associated with the traffic light; receive navigational information indicative of one or more aspects of motion of the first vehicle along the road segment, and determine, based on the navigational information, a lane of travel traversed by the first vehicle along the road segment. The processor may also determine whether the traffic light is relevant to the lane of travel traversed by the first vehicle; update an autonomous vehicle road navigation model relative to the road segment; and distribute the updated autonomous vehicle road navigation model to a plurality of autonomous vehicles.

Abstract: A method for generating labeling data is disclosed that describes an image content of images depicting at least one scene, wherein in a processing device image data are received from an imaging and a segmentation unit that detects at least one object in the image data. A graphical processing unit generates a respective graphical object marker that marks the at least one detected object and a display control unit displays an overlay of the at least one scene and the at least one object marker. An input reception unit receives a respective user input for each object marker, wherein the respective user input provides the image content of the image region marked by the object marker.

Abstract: A lamp controller interlocking system of a camera built-in headlamp inventive concepts includes a headlight module integrated with a camera and a light source, a camera controller generating a single frame image by composing an image captured in a short exposure section, in which a shutter opening time of the camera is relatively short, and an image captured in a long exposure section, in which the shutter opening time of the camera is relatively long, and a lamp controller controlling the light source to emit more light in the long exposure section more than in the short exposure section in synchronization with timings of the long exposure section and the short exposure section of the camera.

Abstract: A device may receive, from a first vehicle, video data for video captured of a location associated with an accident for a second vehicle, and may process the video data, with a first model, to generate a sparse point cloud of the location associated with the accident. The device may process the video data, with a second model, to generate depth maps for frames of the video data, and may utilize the depth maps with the sparse point cloud to generate a dense point cloud. The device may process the video data, with a third model, to generate a dense semantic point cloud, and may process the dense semantic point cloud, with a fourth model, to determine a dense semantic overhead view of the location associated with the accident. The device may perform actions based on the dense semantic overhead view.

Abstract: An image processing device has circuitry, which is configured to obtain image data, the image data being generated on the basis of a non-linear mapping defining a mapping between an object plane and an image plane; and to process the image data by applying a kernel of an artificial network to the image data based on the non-linear mapping.

Abstract: The present disclosure is directed to an autonomous vehicle having a vehicle control system. The vehicle control system includes an image processing system. The image processing system receives an image that includes a plurality of image portions. The image processing system also calculates a score for each image portion. The score indicates a level of confidence that a given image portion represents an illuminated component of a traffic light. The image processing system further identifies one or more candidate portions from among the plurality of image portions. Additionally, the image processing system determines that a particular candidate portion represents an illuminated component of a traffic light using a classifier. Further, the image processing system provides instructions to control the autonomous vehicle based on the particular candidate portion representing an illuminated component of a traffic light.

Abstract: Provided is a system and method for fusion recognition using an active stick filter. The system for fusion recognition using the active stick filter includes a data input unit configured to receive input information for calibration between an image and a heterogeneous sensor, a matrix calculation unit configured to calculate a correlation for projection of information of the heterogeneous sensor, a projection unit configured to project the information of the heterogeneous sensor onto an image domain using the correlation, and a two-dimensional (2D) heterogeneous sensor fusion unit configured to perform stick calibration modeling and design and apply a stick calibration filter.

Abstract: A system (1) for detecting dynamic secondary objects (55) that have a potential to intersect the trajectory (51) of a moving primary object (50), comprising a vision sensor (2) with a light-sensitive area (20) that comprises event-based pixels (21), so that a relative change in the light intensity impinging onto an event-based pixel (21) of the vision sensor (2) by at least a predetermined percentage causes the vision sensor (2) to emit an event (21a) associated with this event-based pixel (21), wherein the system (1) further comprises a discriminator module (3) that gets both the stream of events (21a) from the vision sensor (2) and information (52) about the heading and/or speed of the motion of the primary object (50) as inputs, and is configured to identify, from said stream of events (21a), based at least in part on said information (52), events (21b) that are likely to be caused by the motion of a secondary object (55), rather than by the motion of the primary object (50).

Abstract: A first tone curve referred to in generating an image recognition processing signal as an image output signal for image recognition processing and a second tone curve referred to in generating a visual recognition signal as an image output signal for visual recognition are separately generated. The second tone curve for visual recognition is generated by connecting a dark control point, which is a control point specified in a luminance-pixel value coordinate system corresponding to the lowest luminance portion in a luminance histogram, a light control point, which is a control point specified in the luminance-pixel value coordinate system corresponding to the highest luminance portion in the luminance histogram, and an intermediate control point, which is a control point specified in the luminance-pixel value coordinate system corresponding to an intermediate luminance portion in the luminance histogram.

Abstract: In various examples, a deep neural network (DNN) is trained to accurately predict, in deployment, distances to objects and obstacles using image data alone. The DNN may be trained with ground truth data that is generated and encoded using sensor data from any number of depth predicting sensors, such as, without limitation, RADAR sensors, LIDAR sensors, and/or SONAR sensors. Camera adaptation algorithms may be used in various embodiments to adapt the DNN for use with image data generated by cameras with varying parameters—such as varying fields of view. In some examples, a post-processing safety bounds operation may be executed on the predictions of the DNN to ensure that the predictions fall within a safety-permissible range.

Abstract: A camera module, which is mounted on an inside of a front windshield of a vehicle and configured to image an external environment of the vehicle, includes multiple lens units on which an optical image of the external environment is incident, individually, and an imaging system to generate an outside image of the external environment by imaging through each of the lens units, individually.

Abstract: An image obtaining apparatus, for obtaining an image from an image sensor provided in a vehicle, is provided. The image obtaining apparatus includes: a processor, and a memory including at least one instruction executable by the processor. In response to the at least one instruction being executed by the processor, the processor is configured to: obtain an initial image from the image sensor, the initial image including an object; determine a type of the object included in the initial image; adjust a shutter speed of the image sensor based on the determined type of the object included in the initial image; and obtain a first image from the image sensor based on the adjusted shutter speed.

Abstract: It is an object of the present invention to provide a predictive traffic law enforcement profiler apparatus and method which incorporates a means to determine current location, time, velocity and also incorporates a means to utilize a database derived from historic traffic law enforcement records, crowd sourced records and historical traffic data and also incorporates a predictive processing means to provide historic traffic law enforcement records and estimates of enforced speed limits and enforcement profiles, patrol locations and schedules of traffic law enforcement to a driver.

Abstract: The present disclosure provides a method for pose determination. The method may include obtaining a first pose of a subject at a first time point. The method may further include retrieving one or more first features associated with a road from a database. The road may be viewable by the subject at the first pose and at the first time point. The method may further include obtaining an image captured by the subject at a second time point. The method may further include extracting one or more second features associated with the road from the image. The method may also include determining a second pose of the subject at the second time point at least based on comparing the one or more first features associated with the road with the one or more second features associated with the road.

Abstract: Disclosed is a driving assistance apparatus which can be mounted on a vehicle. The driving assistance apparatus includes an image capturing part, and a processor configured to identify a specific color area in a peripheral image captured by the image capturing part as a first candidate area, identify an area including a specific shape in the first candidate area as a second candidate area, and identify an area of a traffic sign in the second candidate area.

Abstract: Methods, apparatuses, and systems are provided for detecting overhead obstructions along a path segment. One exemplary method includes receiving three-dimensional data collected by a depth sensing device traveling along a path segment, wherein the three-dimensional data comprises point cloud data positioned above a ground plane of the path segment. The method further includes identifying data points of the point cloud data positioned within a corridor positioned above the ground plane. The method further includes projecting the identified data points onto a plane. The method further includes detecting the overhead obstruction based on a concentration of point cloud data positioned within a plurality of cells of the plane. The method further includes storing the detected overhead obstruction above the path segment within a map database.

Abstract: An automated vehicle parking system for a motor vehicle is disclosed and includes at least one camera obtaining images proximate a vehicle, at least one sensor array detecting objects proximate the vehicle, and a controller configured to generate a mask of an area proximate the vehicle from the images obtained by the at least one camera including labeled features identified within the obtained images and project the generated mask onto an occupancy grid generated with information from the at least one sensor array, wherein the controller is configured to locate a parking space responsive to open spaces defined within the generated mask.

Abstract: Machine-learning models are described detecting the signaling state of a traffic signaling unit. A system can obtain an image of the traffic signaling unit, and select a model of the traffic signaling unit that identifies a position of each traffic lighting element on the unit. First and second neural network inputs are processed with a neural network to generate an estimated signaling state of the traffic signaling unit. The first neural network input can represent the image of the traffic signaling unit, and the second neural network input can represent the model of the traffic signaling unit. Using the estimated signaling state of the traffic signaling unit, the system can inform a driving decision of a vehicle.

Abstract: A computing system can receive, in a vehicle, moving object information is determined by processing lidar sensor data acquired by a stationary lidar sensor. The moving object information can be determined using typicality and eccentricity data analysis (TEDA) on the lidar sensor data. The vehicle can be operated based on the moving object information.

Abstract: Systems and methods are disclosed for identifying landmarks. A method for identifying a landmark may include initiating identification of a landmark based on one or more images from a camera, for use in autonomous vehicle navigation, the landmark including a traffic sign; initiating updating a road model with a location of the landmark; and initiating distribution of the road model with the location of the traffic sign to a plurality of autonomous vehicles.

Abstract: A depth image engine and a depth image calculation method are provided. The depth image engine comprises: a buffer, configured to receive a data flow of a first image; an image rotator, configured to perform, after the buffer receives the entire data flow of the first image, a rotation operation on the first image to generate a second image; and a depth value calculator, configured to receive the second image and a reference image to perform a matching calculation on the second image and the reference image to generate a depth image.

Abstract: In various examples, a deep neural network (DNN) is trained—using image data alone—to accurately predict distances to objects, obstacles, and/or a detected free-space boundary. The DNN may be trained with ground truth data that is generated using sensor data representative of motion of an ego-vehicle and/or sensor data from any number of depth predicting sensors—such as, without limitation, RADAR sensors, LIDAR sensors, and/or SONAR sensors. The DNN may be trained using two or more loss functions each corresponding to a particular portion of the environment that depth is predicted for, such that—in deployment—more accurate depth estimates for objects, obstacles, and/or the detected free-space boundary are computed by the DNN. In some embodiments, a sampling algorithm may be used to sample depth values corresponding to an input resolution of the DNN from a predicted depth map of the DNN at an output resolution of the DNN.

Abstract: A parallel computing method for man-machine coordinated steering control of a smart vehicle based on risk assessment is provided, comprising the following steps: building a lateral kinetic equation model of a vehicle; building a target function by targeting at minimizing an offset distance of a vehicle driving track from a lane center line and making a change in a front wheel steering angle and a longitudinal acceleration as small as possible in a driving process; building a parallel computing architecture of a prediction model and the target function, and employing a triggering parallel computing method; solving and computing a gradient with a manner of back propagation and using a gradient descent method to obtain an optimal control amount of the front wheel steering angle and an optimal control amount of the longitudinal acceleration; and computing a driving weight, obtaining a desired front wheel steering angle and completing real time control.

Abstract: Techniques including receiving a distorted image from a camera disposed about a vehicle, detecting, in the distorted image, corner points associated with a target object, mapping the corner points to a distortion corrected domain based on one or more camera parameters, mapping the corner points and lines between the corner points back to a distorted domain based on the camera parameters, interpolating one or more intermediate points to generate lines between the corner points in the distortion corrected domain mapping the corner points and the lines between the corner points back to a distorted domain based on the camera parameters, and adjusting a direction of travel of the vehicle based on the located target object.

Abstract: The disclosure is directed to systems and methods for operating or controlling an ADAS mechanism. A first sensor of an ADAS mechanism can determine a potential obstacle to a vehicle, and a position of the potential obstacle relative to the first sensor. A second sensor can determine a gaze angle of a user of the vehicle. An activation engine in communication with the first sensor and the second sensor can determine a proximity of the gaze angle of the user to the determined position of the potential obstacle. The activation engine can control, in response to the determined proximity, an operation of the ADAS mechanism for responding to the potential obstacle.

Abstract: A method of operating a vehicle of determining whether the vehicle is operating in a road segment with a low visibility condition to cause a loss of input of sensor data to a vehicle controller that operates an assist feature, activating one or more adaptive alerts based on a road departure risk of the vehicle, and driver use of the assist feature in the upcoming road segment, wherein the road departure risk is determined by calculating a road departure risk index that compares an estimated vehicle path based on the vehicle state data with a probabilistic vehicle path for the upcoming road segment; and predicting whether will operate within an acceptable path in the upcoming road segment; and tracking the vehicle in the upcoming road segment based on vehicle navigation data to provide at least one adaptive alert based on a prediction of the road departure risk.