Abstract: An excavator includes a rotatable house and a bucket operably coupled to the house. An inertial measurement unit (IMU) is operably coupled to the excavator and is configured to provide at least one IMU signal indicative of rotation of the house. A backup camera is disposed to provide a video signal relative to an area behind the excavator. A controller is coupled to the IMU and operably coupled to the backup camera. The controller is configured to receive the at least one IMU signal from the IMU and to generate a position output based on the at least one IMU signal and the video signal from the backup camera.

Abstract: Provided is a driving apparatus including a body, a surroundings detection unit detecting surroundings of the body, the surrounding detection unit including a light detection and ranging (LIDAR) device and a camera, and at least one processor configured to control the driving of the body based on a detection result from the surroundings detection unit, wherein the at least one processor is further configured to detect a ramp and flat ground that are present on a driving path based on an image obtained by the camera, and determine the detected ramp or the detected flat ground from a detection result from the LIDAR device as a non-obstacle.

Abstract: The present disclosure relates to a computer system, configured to receive two or more instances of a first route object comprising a geotemporal attribute comprising geospatial and temporal data to define an instance of the first route object; map the two or more instances of the first route object on to digital map data to define respective first route vertices in the digital map; and receive user input defining an interval and applying the defined interval intermediate chronologically adjacent respective first route vertices to define first waypoints between the respective first route vertices.

Abstract: A vehicle hitching assistance system includes a controller executing an initial alignment process including searching for a trailer within a specified area of image data that is less than a total field of the image data in directions corresponding with both a longitudinal distance between the vehicle and the trailer and a lateral direction perpendicular to the longitudinal distance, presenting an indication to a driver of the vehicle to reposition the vehicle when the trailer is not identified within the specified area, and removing the indication when the trailer is identified within the specified area. The controller further executes an automated backing process upon identifying the trailer within the specified area, including identifying a coupler of the trailer and outputting a steering signal to the vehicle to cause the vehicle to steer to align a hitch ball of the vehicle with the coupler of the trailer during reversing.

Abstract: In various examples, a machine learning model—such as a deep neural network (DNN)—may be trained to use image data and/or other sensor data as inputs to generate two-dimensional or three-dimensional trajectory points in world space, a vehicle orientation, and/or a vehicle state. For example, sensor data that represents orientation, steering information, and/or speed of a vehicle may be collected and used to automatically generate a trajectory for use as ground truth data for training the DNN. Once deployed, the trajectory points, the vehicle orientation, and/or the vehicle state may be used by a control component (e.g., a vehicle controller) for controlling the vehicle through a physical environment. For example, the control component may use these outputs of the DNN to determine a control profile (e.g., steering, decelerating, and/or accelerating) specific to the vehicle for controlling the vehicle through the physical environment.

Abstract: A vehicular safety feature control system includes a camera disposed at a vehicle that views exterior of the equipped vehicle. The vehicular safety feature control system includes an electronic control unit (ECU) for processing image data captured by the camera to detect presence of objects. The ECU, responsive to processing of image data captured by the camera, detects an object on a collision course with the equipped vehicle. The ECU, responsive to detecting the object on the collision course with the equipped vehicle, classifies the detected object. The ECU estimates a mass of the detected object based on the classification. The ECU estimates a kinetic energy of a potential collision and, based on the estimated kinetic energy of the potential collision, estimates a severity of the potential collision. The ECU controls a safety feature of the equipped vehicle based on the estimated severity of the potential collision.

Abstract: Disclosed are a track mergence module and a method by which track deletion conditions are set to be varied depending on various situations and a track corresponding to the set track deletion conditions is deleted. The track mergence module includes a track deletion condition generation and storage unit, a phenomenon determination unit and a track mergence unit. A track to be deleted may be accurately deleted by generating the track deletion conditions so as to adaptively correspond to the performance of a sensor and various situations in which a host vehicle is placed, thereby being capable of erroneous braking of the autonomously driving host vehicle.

Abstract: Enclosed are embodiments for sampling driving scenarios for training machine learning models. In an embodiment, a method comprises: assigning, using at least one processor, a set of initial physical states to a set of objects in a map for a set of simulated driving scenarios, wherein the set of initial physical states are assigned according to one or more outputs of a random number generator; generating, using the at least one processor, the set of simulated driving scenarios in the map using the initial physical states of the objects in the set of objects; selecting, using the at least one processor, samples of the simulated driving scenarios; training, using the at least one processor, a machine learning model using the selected samples; and operating, using a control circuit, a vehicle in an environment using the trained machine learning model.

Abstract: An unmanned device used in a wireless communication relay system includes: a surrounding environment information detector acquiring information on a surrounding environment of a moving object; a processor generating a control command for the moving object based on the acquired information on the surrounding environment; and a communication interface for transmitting the control command generated by the processor to the moving object, wherein the unmanned device is located at a predetermined height and relays communication between the moving object and a control center, and thus, even when wireless communication between the control center and the moving object is difficult due to an obstacle or the like, the connection of the wireless communication may be maintained without disconnection of the wireless communication.

Abstract: A control system for a wheel loader includes an upper sensor installed in a driver cabin to obtain shape information data for an object in front of the driver cabin, a lower sensor installed in a front body to obtain shape information data for an object in front of the front body, a work apparatus position detection portion configured to detect a position of a work apparatus connected rotatably to the front body, and an obstacle detection control device configured to receive the shape information data from the upper sensor and the lower sensor and configured to calculate a distance to the object based on the information data of any one selected from the upper sensor and the lower sensor according to the detected position of the work apparatus.

Abstract: A steering control device includes an electronic control unit. The electronic control unit is configured to acquire a limit position determination angle corresponding to an absolute steering angle when it is determined that movement of a turning shaft is limited, to compare a first stroke width which is a sum of an absolute value of the limit position determination angle on a right side and an absolute value of the limit position determination angle on a left side with a stroke threshold value corresponding to an entire stroke range of the turning shaft when the limit position determination angles on the right and left sides are acquired, and to set end-position-corresponding angles on the right and left sides based on the limit position determination angles on the right and left sides when the first stroke width is greater than the stroke threshold value.

Abstract: In some implementations, a device may receive video data associated with video frames that depict an environment of a vehicle. The device may identify an object depicted in the video frames, wherein an object detection model indicates bounding boxes associated with the object. The device may determine, based on the bounding boxes, a configuration of the bounding boxes within the video frames. The device may determine, based on the configuration of the bounding boxes, that the object is a vulnerable road user (VRU) that is in the environment. The device may determine a trajectory of the VRU based on a change in the configuration between video frames of a set of the video frames. The device may determine, based on the trajectory, a probability of a collision between the VRU and the vehicle. The device may perform, based on the probability, an action associated with the vehicle.

Abstract: A machine may receive, from one or more sensors, first data indicating a first position of a work implement within an environment and second data indicating a second position of the work implement within the environment. Based on the first position and the second position, the machine may define a working area of the work implement. The machine may receive, from one or more cameras, image data depicting at least a portion of the environment and at least a portion an object disposed within a field of view of the one or more cameras. The machine may determine, based on the image data, that the object is located outside of the working area. Based on determining that the object is located outside of the working area, the machine may refrain from causing output of an alert indicating the presence of the object outside the working area.

Abstract: Various technologies described herein pertain to autonomous vehicle consumption of real-time public transportation data to guide curb access and usage. An autonomous vehicle receives a trip request for a ride specifying a requested pullover location. The autonomous vehicle receives public transportation data specifying an expected arrival time of a public transportation vehicle at a reserved zone within proximity of the requested pullover location. The autonomous vehicle evaluates availability of the reserved zone during an expected occupancy time of the reserved zone by the autonomous vehicle based on the expected arrival time of the public transportation vehicle at the reserved zone. The autonomous vehicle selects an actual pullover location for the ride in the autonomous vehicle based on the availability of the reserved zone during the expected occupancy time. The autonomous vehicle stops at the actual pullover location for the ride in the autonomous vehicle.

Abstract: A machine-learning (ML) architecture for determining three or more outputs, such as a two and/or three-dimensional region of interest, semantic segmentation, direction logits, depth data, and/or instance segmentation associated with an object in an image. The ML architecture may output these outputs at a rate of 30 or more frames per second on consumer grade hardware.

Abstract: Systems and methods for monitoring and analyzing vehicle data within a vehicle and providing analytical processing data to prospective users of vehicles are disclosed. In one embodiment, a method is disclosed comprising monitoring a communications bus installed within a vehicle, the communications bus transmitting data recorded by one or more sensors installed within the vehicle; detecting a message broadcast on the communications bus; extracting an event from the message, the extraction based on a pre-defined list of event types; storing the event in a secure storage device installed within the vehicle; determining that a transfer condition has occurred; and transferring the event data to a remote server in response to determining that the transfer condition has occurred.

Abstract: Techniques for increasing performance of machine-learned models while conserving computational resources generally required by ensemble machine-learning methods are described herein. The techniques may include determining multiple views of a scene that is to be input into a machine-learned model. In some examples, a scene data input may be rotated by 90, 180, and 270 degrees to generate four scene inputs (e.g., 0-, 90-, 180-, and 270-degree rotated inputs) that can be passed through the machine-learned model and the results per scene can be aggregated to determine a final prediction/decision. Similarly, scene inputs may be shifted, reflected, translated, and/or the like before being input into the machine-learned model. The predictions may be associated with one or more objects in the environment that are represented in the scenes.

Abstract: The invention relates to a system and method for navigating an autonomous driving vehicle (ADV) that utilizes an-onboard computer and/or one or more ADV control system nodes in an ADV network platform. The on-board computer receives battery monitoring and management data concerning a battery stack. The on-board computer, utilizing a battery management system, determines the current state of charge (SOC) and other information concerning the battery stack and determines if the estimated total amount of electrical power required to navigate an ADV along a generated route to reach the predetermined destination is available. In response to determining that the ADV cannot reach the predetermined destination, the on-board computer automatically initiates a dynamic routing algorithm, which utilizes artificial intelligence, to generate alternative routes in an effort to find a route that the ADV can navigate to reach the destination utilizing the current state of charge (SOC) of the battery stack.

Abstract: A monitoring system and method are provided to monitor ground movement of aircraft driven with electric taxi drive systems and movement of ground service vehicles and equipment and personnel within airport ramp areas. Monitor and sensor devices, including those that are intelligent and employ scanning technology to generate image, positional, and other data may be mounted in locations on aircraft, ground service vehicles and equipment, passenger loading bridges, an airport terminal, and other ramp locations to generate a constant stream of data as the aircraft moves into, within, and out of an airport ramp area. The data stream is transmitted to an artificial intelligence-based processing system to identify and communicate possible safety hazards to multiple locations so that the aircraft's ground travel or a ground vehicle's travel may be altered to avoid identified safety hazards and to avoid collisions.

Abstract: A vehicle can include an on-board data processing system that receives sensor data captured by various sensors of the vehicle. As a vehicle travels along a route, the on-board data processing system can process the captured sensor data to identify a potential vehicle stop. The on-board data processing system can then identify geographical coordinates of the location at which the potential vehicle stop occurred, use artificial intelligence to classify a situation of the vehicle at the potential stop, and determine whether the stop was caused by a road obstacle, such as a speed bump, a gutter, an unmarked crosswalk, or any other obstacle not at an intersection. If the stop was caused by the road obstacle, the on-board data processing system can generate virtual stop or yield line data corresponding to the identified geographic coordinates and transmit this data to a server over a network for processing.