Abstract: Disclosed are a method and an apparatus for generating a map for autonomous driving and recognizing a location based on the generated map. When generating a map, a spherical range image is obtained by projecting 3D coordinate information corresponding to a 3D space onto a 2D plane, and semantic segmentation is performed on the spherical range image to generate a semantic segmented image. Then, map data including a spherical range image, a semantic segmented image, and lane attribute information are generated.

Abstract: Disclosed herein are a method and apparatus for processing a driving cooperation message. The method for processing a driving cooperation message includes receiving multiple first driving cooperation messages from neighboring autonomous vehicles, adjusting cooperation classes of the multiple first driving cooperation messages, creating driving strategies corresponding to the adjusted cooperation classes in descending order of priorities of the adjusted cooperation classes, generating second driving cooperation messages including the adjusted cooperation classes and the driving strategies corresponding to the adjusted cooperation classes, and sending the second driving cooperation messages to the neighboring autonomous vehicles requiring cooperative driving.

Abstract: Disclosed herein are an apparatus and method for adaptive autonomous driving control. The apparatus includes memory in which at least one program is recorded and a processor for executing the program. The program may perform control of a target vehicle by converting a theoretical control value based on a vehicle control algorithm into a hardware-dependent control value, which is dependent on the platform or hardware of the target vehicle, and may modify at least one parameter or a conversion equation for conversion of the hardware-dependent control value such that an error is minimized based on the difference between a response value according to the control of the target vehicle and a control value.

Abstract: The present disclosure relates to a control apparatus and method of a rear-wheel steering system. The control apparatus of a rear-wheel steering system includes: a plurality of sensors configured to detect information on a vehicle status; and a processor configured to: check the vehicle status in response to signals input from the plurality of sensors and determine a control mode for a rear-wheel steering (RWS) system as one of reverse-phase control, in-phase control, and neutral control; and, in the event of a sharp steering situation, change the control mode for the RWS system and perform control. The control apparatus and method of a rear-wheel steering system may improve responsiveness through reverse-phase control in the RWS system and enhance safety by performing neutral control in a sharp steering situation, thereby enhancing steering performance of the RWS system.

Abstract: The present invention relates to a system for eliminating noise of a LiDAR sensor caused by an adverse weather environment and an operating method thereof. An operating method of a light detection and ranging (LiDAR) sensor noise elimination system includes receiving a first point cloud containing noise from a LiDAR sensor, generating a two-dimensional first range image and a two-dimensional first reflectance image based on the first point cloud, generating a two-dimensional first noise region boundary image by inputting the first range image and the first reflectance image into a pre-trained machine learning model, and eliminating noise contained in the first point cloud using the first noise region boundary image.

Abstract: Disclosed are an apparatus and method for controlling traffic lights, and more particularly, an apparatus and method for controlling traffic lights, which can provide information on the remaining time of a traffic signal. The apparatus for controlling traffic lights includes a traffic signal display controller configured to transmit a pattern control command including information related to the remaining time of a current traffic signal and traffic lights configured to display information on a traffic signal by receiving the pattern control command.

Abstract: Disclosed herein is a method for deidentifying a driver image dataset. The method includes generating a combination dataset having a preset size based on a driver image dataset, extracting face shape information from each of pieces of driver image data forming the driver image dataset, and generating a deidentified dataset using the combination dataset and the face shape information.

Abstract: A training data service system for operation scope-oriented autonomous driving shuttle includes a communication module configured to collect an autonomous driving data set from an autonomous driving vehicle to construct training data and distribute the training data to an autonomous driving shuttle, a memory configured to store a program for construction of the training data, and a processor configured to generate query data including road shape information, road attribute information, traffic environment information, and collection sensor information by executing the program stored in the memory, detect data satisfying a condition corresponding to the query data from a previously collected autonomous driving data set, construct training data for the autonomous driving shuttle, based on the detected data, and distribute the constructed training data to a corresponding autonomous driving shuttle.

Abstract: The present invention relates to a method and apparatus for predicting future trajectories of nearby vehicles for autonomous driving. The method of predicting the future trajectories of nearby vehicles according to the present invention includes generating past trajectories of the nearby vehicles, extracting local routes along which the nearby vehicles are travelable from a high-definition map, encoding the past trajectories and the local routes, and generating the future trajectories of the nearby vehicles using a deep neural network on the basis of the encoded past trajectories and an updated local route encoding.

Abstract: According to the present invention, autonomous driving may be performed according to an autonomous driving strategy defined in trajectory constraints generated based on object recognition information about a road map and various traffic participants, and thus, because collision avoidance does not obstruct traffic of an adjacent lane despite a situation where an autonomous vehicle stops for avoiding a collision, the stability of autonomous driving may increase and vehicle traffic management may be easily performed.

Abstract: Provided are a method and system for generating a destination for an emergency response of an autonomous vehicle of an autonomous driving system. A method of generating a destination of an autonomous vehicle according to the present invention includes generating forward perception information based on data collected from a sensor mounted on the autonomous vehicle, setting a destination generation area based on the forward perception information, generating a candidate destination in the destination generation area based on the forward perception information and a current heading range of the autonomous vehicle, and when the candidate destination is provided as a plurality of candidate destinations, selecting one destination from among the candidate destinations based on a maximum vertical movement distance of each of the candidate destinations.

Abstract: A precise position correction method based on collaborative cognition in an autonomous vehicle includes: estimating object information on a road within a detection range on an autonomous vehicle; receiving a shared message from at least one of a first vehicle (another connected vehicle (CV) or automated vehicle (CAV)) and a road infrastructure device for V2X communication on the road, receiving the message including driving status information of the other CV when the first vehicle is the other CV, and receiving the driving status information of the other CAV and the message about a second vehicle, in which the V2X communication is not possible, recognized by the other CAV; calculating an estimated error by comparing the estimated object information with the message; and setting a radius based on the first vehicle corresponding to an estimated error application radius in consideration of a distance between the autonomous vehicle and the first vehicle.

Abstract: Disclosed herein are a deep network learning method using an autonomous vehicle and an apparatus for the same. The deep network learning apparatus includes a processor configured to select a deep network model requiring an update in consideration of performance, assign learning amounts for respective vehicles in consideration of respective operation patterns of multiple autonomous vehicles registered through user authentication, distribute the deep network model and the learning data to the multiple autonomous vehicles based on the learning amounts for respective vehicles, and receive learning results from the multiple autonomous vehicles, and memory configured to store the deep network model and the learning data.

Abstract: Disclosed are a device for multiple interactions between moving objects and autonomous vehicles and a driving method thereof. The device for multiple interactions between moving objects and autonomous vehicles includes memory in which a program for multiple interactions between moving objects and autonomous vehicles has been stored and a processor configured to execute the program. The processor confirms the status of the moving object including at least any one of a nearby vehicle, a pedestrian, and a mobility user and transmits a control command for a notification function that shares situation information.

Abstract: A method of filtering dynamic objects in radar-based ego-motion estimation includes converting measurement value at current time, measured by radar sensor, into point cloud, classifying the point cloud into points of a first object predicted as static object and points of a second object predicted as dynamic object, based on position value of dynamic object tracked at previous time, classifying the points of the first object into the points of the static object predicted as normal value and the points of the dynamic object predicted as outlier, based on outlier filtering algorithm, classifying the points of the second object into points of a candidate static object and points of a candidate dynamic object, based on velocity model of the static object, and tracking a position value of the dynamic object at current time, based on the points of the dynamic object and the points of the candidate dynamic object.

Abstract: Disclosed is a processor which includes a camera image feature extractor that extracts a camera image feature based on a camera image, a LIDAR image feature extractor that extracts a LIDAR image feature based on a LIDAR image, a sampling unit that performs a sampling operation based on the camera image feature and the LIDAR image feature and generates a sampled LIDAR image feature, a fusion unit that fuses the camera image feature and the sampled LIDAR image feature and generates a fusion map, and a decoding unit that decodes the fusion map and generates a depth map. The sampling operation includes back-projecting a pixel location of the camera image feature on a camera coordinate system to generate a back-projection point, and projecting the back-projection point on a plane of the LIDAR image to calculate sampling coordinates.

Abstract: A method of predicting a possibility of an accident is provided. The method includes abstracting surrounding situation data and movement data of an ego-vehicle input from a sensor to generate abstracted driving situation data by using an abstraction module executed by a processor, calculating a digitized score of a possibility of an accident of the ego-vehicle by using a calculation module executed by the processor, based on the abstracted driving situation data, and generating action data of the ego-vehicle for decreasing the possibility of the accident by using an action generating module executed by the processor, based on the score.

Abstract: Disclosed is a processor which includes a camera image feature extractor that extracts a camera image feature based on a camera image, a LIDAR image feature extractor that extracts a LIDAR image feature based on a LIDAR image, a sampling unit that performs a sampling operation based on the camera image feature and the LIDAR image feature and generates a sampled LIDAR image feature, a fusion unit that fuses the camera image feature and the sampled LIDAR image feature and generates a fusion map, and a decoding unit that decodes the fusion map and generates a depth map. The sampling operation includes back-projecting a pixel location of the camera image feature on a camera coordinate system to generate a back-projection point, and projecting the back-projection point on a plane of the LIDAR image to calculate sampling coordinates.

Abstract: A method of filtering dynamic objects in radar-based ego-motion estimation includes converting measurement value at current time, measured by radar sensor, into point cloud, classifying the point cloud into points of a first object predicted as static object and points of a second object predicted as dynamic object, based on position value of dynamic object tracked at previous time, classifying the points of the first object into the points of the static object predicted as normal value and the points of the dynamic object predicted as outlier, based on outlier filtering algorithm, classifying the points of the second object into points of a candidate static object and points of a candidate dynamic object, based on velocity model of the static object, and tracking a position value of the dynamic object at current time, based on the points of the dynamic object and the points of the candidate dynamic object.

Abstract: A method for changing a route when an error occurs in an autonomous driving AI includes collecting error information of the AI when an error of the AI has occurred, extracting, from a storage, past error information about a same kind of AI as that of the AI based on the error information of the AI, generating an error analysis result based on the past error information, generating an error analysis result message based on the error analysis result, and determining whether the driving of the autonomous driving vehicle needs to be stopped based on the error analysis result message.