An apparatus comprises a plurality of sensors, a digital map. and a control unit. The plurality of sensors may be configured to detect information about an exterior environment of a vehicle. The digital map may be configured to provide information about roadways in a vicinity of the vehicle. The control unit (i) may comprise an interface configured to receive (a) sensor status signals, (b) sensor-based information, and (c) map-based information, and (ii) may be configured to (a) determine whether an operational situation exists that is unsafe for an advanced driver-assistance systems (ADAS) automation feature to be activated or remain active based on the sensor-based information, the map-based information, and the sensor status signals, and (b) generate an activation control signal to restrict activation of the ADAS automation feature when an unsafe operational situation exists.

A collision warning method in relation to vulnerable road users (VRUs), applied in a host vehicle when being driven, obtains a relative lateral distance between a vehicle and a VRU and obtains speed of movement of the VRU in addition to speed and direction of the host vehicle. A warning scenario is determined according to the relative lateral distance and the VRU speed. A warning distance according to the warning scenario and the relative speed is calculated, a relative distance between the host vehicle and the VRU is obtained, and the issue of a corresponding level of a collision warning is determined according to a comparison between the warning distance and the relative distance.

A recognition processing apparatus includes a video acquisition unit configured to acquire first captured image data of surroundings of a host vehicle captured by a far-infrared camera, an other vehicle detection unit configured to detect another vehicle parked or stopped in the surroundings of the host vehicle, a heat detection unit configured to detect radiated heat associated with an operation of the other vehicle based on a thermal distribution in a region corresponding to the other vehicle in the first captured image data, and a person detection unit configured to, when the radiated heat has been detected, preferentially execute person recognition in a vicinity of the other vehicle to detect a person. The recognition processing apparatus can ascertain the possibility that a person gets out of a detected vehicle and promptly detect such a person.

Low variance detection training is described herein. In an example, annotated data can be determined based on sensor data received from a sensor associated with a vehicle. The annotated data can comprise an annotated low variance region and/or an annotated high variance region. The sensor data can be input into a model, and the model can determine an output comprising a high variance output and a low variance output. In an example, a difference between the annotated data and the output can be determined and one or more parameters associated with the model can be altered based at least in part on the difference. The model can be transmitted to a vehicle configured to be controlled by another output of the model.

Systems and methods for lane marking and road sign recognition are provided. The system aligns image level features between a source domain and a target domain based on an adversarial learning process while training a domain discriminator. The target domain includes one or more road scenes having lane markings and road signs. The system selects, using the domain discriminator, unlabeled samples from the target domain that are far away from existing annotated samples from the target domain. The system selects, based on a prediction score of each of the unlabeled samples, samples with lower prediction scores. The system annotates the samples with the lower prediction scores.

The disclosure includes embodiments for pedestrian navigation by a group of connected vehicles executing a collaborative computing process. In some embodiments, a method includes analyzing pedestrian data generated by a pedestrian device and sensor data generated by the group of connected vehicles to determine digital twin data from a set that correlates with a scenario described by the pedestrian data and the sensor data. The digital twin data is an output of a historical digital twin simulation. The method includes predicting, based on the digital twin data, that the pedestrian is at risk of a collision. The method includes determining modified path data describing a modified walking path for the pedestrian. The method includes transmitting the modified path data to the pedestrian device so that the pedestrian is informed about the modified walking path and the risk is modified.

System, methods, and other embodiments described herein relate to a training system to train a driver about occurrences of anomalous driving events of automated vehicle systems. In one embodiment, a method includes determining, upon receiving a selection of a vehicle behavior from one or more anomalous driving events and a detected state change signal, whether the vehicle behavior affects one or more entities. The method includes assessing a state of the one or more entities to simulate the vehicle behavior according to a safety standard. The method includes triggering simulation of the vehicle behavior if the state satisfies a threshold. The method includes simulating the vehicle behavior by at least controlling the vehicle to simulate the vehicle behavior during automated driving of the vehicle.

A vehicle exit assist apparatus includes a target information acquisition device configured to detect a target present around a vehicle and acquire information regarding the detected target as target information; and a control unit configured to determine whether or not an obstruction target having a possibility of obstructing a safe vehicle exit of an occupant of the vehicle while the vehicle is stopped is present based on the target information, and to execute vehicle exit assist control that assists in the safe vehicle exit of the occupant when determination is made that the obstruction target is present. The control unit is configured to determine whether or not the obstruction target is a human being, and not to execute the vehicle exit assist control when determination is made that the obstruction target is a human being.

Techniques for controlling a vehicle based on height data and/or classification data being determined utilizing multi-channel image data are discussed herein. The vehicle can capture lidar data as it traverses an environment. The lidar data can be associated with a voxel space as three-dimensional data. Semantic information can be determined and associated with the lidar data and/or the three-dimensional voxel space. A multi-channel input image can be determined based on the three-dimensional voxel space and input into a machine learned (ML) model. The ML model can output data to determine height data and/or classification data associated with a ground surface of the environment. The height data and/or classification data can be utilized to determine a mesh associated with the ground surface. The mesh can be used to control the vehicle and/or determine additional objects proximate the vehicle.

Systems and methods for assisting navigation of a vehicle are disclosed. In one embodiment, a method of assisting navigation of a vehicle includes receiving navigational data relating to an intended route of the vehicle, receiving object data relating to at least one external object detected within a vicinity of a current position of the vehicle, determining whether the at least one external object affects an ability of the vehicle to proceed along the intended route, and generating at least one instruction relating to the ability of the vehicle to proceed along the intended route.