The present invention relates to a method for operating a driver model for controlling a vehicle. The driver model comprises a vehicle module (203) which determines an accelerator pedal position to be set on the vehicle. In addition, the vehicle module (203) determines a required power as a component of a total power, which total power can be generated by a drive system of the vehicle, wherein the required power corresponds to a power that is necessary for moving the vehicle at a required speed and/or a required acceleration (311) along a predefined road course. The method according to the invention further provides for a value (313) of a permissible pedal position to be assigned to the required power and for the value (313) of the permissible pedal position to be transmitted to the driver model in order to control the vehicle.

A method of controlling autonomous driving is provided. The method includes collecting driving information of a driver and curvature information of a road and generating a driving pattern of the driver, defined by associating behaviors in longitudinal and lateral directions of the vehicle based on the driving information. The driving pattern is then set to a constraint condition for driving torque and brake pressure and the vehicle is operated based thereon.

A method is provided for driving model calibration. The method clusters a plurality of vehicle trajectories into a plurality of datasets for different driving styles based on a score. The score is calculated for each vehicle trajectory by an objective entropy weight method. The method trains, for each of the plurality of datasets for the different driving styles relative to an existing target driving model, a respective neural network which inputs a respective one of the plurality of datasets and outputs a respective parameter for the existing target driver model to obtain a plurality of trained neural networks. The existing target driver model is for simulating human driving behaviors. The method performs, for each trained neural network, an online adaptation of the existing target driving model based on a respective output of each of the plurality of trained neural networks to obtain a plurality of adapted driver models.

The present teaching relates to method, system, and medium, for operating a vehicle. Real-time data related to the vehicle are received. A current mode of operation of the vehicle and a state of the driver present in the vehicle are determined. A first risk associated with the current mode of operation of the vehicle is evaluated based on the real-time data and the state of the driver in accordance with a risk model. In response to the first risk satisfying a first criterion, a second risk associated with switching the current mode to a different mode of operation of the vehicle is determined based on the state of the driver. The vehicle is switched from the current mode to the different mode when the second risk satisfies a second criterion.

A method for adaptively optimizing an autonomous driving system includes obtaining a driving intent of a driver of a target vehicle, wherein the target vehicle is an autonomous driving vehicle controlled by the autonomous driving system, detecting, based on the driver's driving intent, that a conflict exists between a first driving behavior of the target vehicle controlled by the autonomous driving system and the driver's driving intent, and updating the autonomous driving system such that a driving behavior of the target vehicle controlled by an updated autonomous driving system matches the driving intent.

A distracted driving analysis system for identifying distracted driving events is provided. The system includes a processor in communication with a memory device programmed to: (i) receive driving event records, each driving event record including phone usage by a user, wherein a driving event record is labeled as an actual distracted driving event or a passenger event, (ii) divide the driving event records into at least two clusters based at least in part upon common features and the labels of each driving event record by processing the plurality of driving event records with a semi-supervised machine learning algorithm, (iii) generate a trained model based at least in part upon the at least two clusters, (iv) process a new driving event using the trained model, (v) assign the new driving event to one of the clusters using the trained model, and/or (vi) determine whether the new driving event is an actual distracted driving event or a passenger event.

A method for producing a model (15) for automated prediction of interactions of a user with a user interface of a motor vehicle. Vehicle operating logs (11, 12, 13) are provided and each includes a record of a time sequence of user interactions with the user interface. Each of the user interactions recorded in the vehicle operating logs (11, 12, 13) is assigned context information (21, 22) that includes a functional category (21) of the user interaction and a driving state (22) of the motor vehicle at the time of the user interaction. Training data (14) are generated based on the vehicle operating logs (11, 12, 13) and the associated context information (21, 22). A context-sensitive interaction model (15) is trained by machine learning on the basis of the training data (14) to make a prediction about a future user interaction based on a time sequence of past user interactions.

In various embodiments, a driver sensing subsystem computes a characterization of a driver based on physiological attribute(s) of the driver that are measured as the driver operates a vehicle. Subsequently, a driver assessment application uses a confidence level model to estimate a confidence level of the driver based on the characterization of the driver. The driver assessment application then causes driver assistance application(s) to modify at least one functionality of the vehicle based on the confidence level. Advantageously, by enabling the driver assistance application(s) to take into account the confidence level of the driver, the driver assessment application can improve driving safety relative to conventional techniques for implementing driver assistance applications that disregard the confidence levels of drivers.

The present teaching relates to method, system, medium, and implementation of human-like vehicle control for an autonomous vehicle. Information related to a target motion to be achieved by the autonomous vehicle is received, wherein the information includes a current vehicle state of the autonomous vehicle. A first vehicle control signal is generated with respect to the target motion and the given vehicle state in accordance with a vehicle kinematic model. A second vehicle control signal is generated in accordance with a human-like vehicle control model, with respect to the target motion, the given vehicle state, and the first vehicle control signal, wherein the second vehicle control signal modifies the first vehicle control signal to achieve human-like vehicle control behavior.

A system and method for path planning of autonomous vehicles based on gradient are disclosed. A particular embodiment includes: generating and scoring a first suggested trajectory for an autonomous vehicle; generating a trajectory gradient based on the first suggested trajectory; generating and scoring a second suggested trajectory for the autonomous vehicle, the second suggested trajectory being based on the first suggested trajectory and a human driving model; and outputting the second suggested trajectory if the score corresponding to the second suggested trajectory is within a score differential threshold relative to the score corresponding to the first suggested trajectory.