Systems and methods for controlling navigation of an autonomous vehicle through an intersection are disclosed. The methods include determining a loiter pose of an autonomous vehicle for stopping at a point within the intersection before initiating an unprotected turn for traversing the intersection. One or more distinct classes of trajectories are then identified, each of which is associated with multiple trajectories that take the same combination of discrete actions with respect to the loiter pose. A constraint set for each of the one or more distinct classes of trajectories is then be computed based on the loiter pose, and a candidate trajectory is determined for each of the one or more distinct classes based on the corresponding constraint set. A trajectory for the autonomous vehicle for executing the unprotected turn for traversing the intersection is selected from amongst the candidate trajectories.

A risk prediction determination device includes a predicted traveling route specifying unit, a risk location specifying unit, and a risk prediction determination unit. The predicted traveling route specifying unit specifies a predicted traveling route of a subject vehicle based on traveling lane information indicating a traveling lane of the subject vehicle, road type information indicating a type of a road around the subject vehicle, road link information related to a road link, and sensor information of the subject vehicle. The risk location specifying unit specifies a location with risk based on road shape information around the subject vehicle and location information of an obstacle. The risk prediction determination unit performs a risk prediction determination based on a specified result of the predicted traveling route specifying unit and a specified result of the risk location specifying unit.

According to one aspect, uncertainty prediction based deep learning may include receiving, using a memory, a trained neural network policy π trained based on a first dataset in a first environment, implementing, via a controller, the trained neural network policy π in a second environment by receiving an input and generating an output y, calculating an uncertainty array U[T] for a time window T, wherein the uncertainty array is indicative of a level of uncertainty associated with an output sample distribution of the output across the time window T based on a temporal divergence, an entropy H, a variational ratio VR, and a standard deviation SD of the output y, and executing, via the controller and one or more systems, an action based on the uncertainty array U[T], such as discontinuing use of the trained neural network policy π.

A vehicle control system for a semi-autonomous vehicle is provided. The vehicle control system includes a controller coupled to a plurality of sensors positioned within the vehicle and to a heads-up display (HUD). The controller includes a processor in communication with a memory device. The controller receives sensor data from the plurality of sensors, determines the vehicle is turning, identifies, based on the sensor data, a candidate turn path for the vehicle, and identifies an actual turn path for the vehicle. The controller also transmits, to one or more automation systems of the vehicle, a control signal that instructs the automation systems to perform a turn-assist function to reduce a determined deviation between the actual turn path and the candidate turn path and to transmit, to the HUD, a control signal that instructs the HUD to display a notification to a driver of the vehicle of the turn-assist function.

Systems, methods and computer-readable media for selecting a trajectory for an autonomous vehicle are disclosed that include computing a current vehicle state for the autonomous vehicle based on observations by a sensing system; computing respective collision probability scores for a plurality of candidate trajectories based on the current vehicle state; computing respective information gain scores for the plurality of candidate trajectories based on the current vehicle state, the information gain score for each candidate trajectory indicating an respective information gain for a next planning horizon interval that is subsequent to the current planning horizon interval; and selecting a planned trajectory from the plurality of candidate trajectories based on the respective collision probability scores and respective information gain scores.

A method of adapting tuning parameter settings of a system (2) functionality (3) for road vehicle (1) speed adjustment control starting from initially selected settings and applying a training set of speed adjustment profiles obtained from manually negotiated road segments and road segment data for these. For each of these road segments: —a simulated speed adjustment profile is calculated using the selected settings and the road segment data; —the manual and the simulated speed adjustment profiles are compared to obtain a residual; —a norm of the residual is calculated. For all of the road segments of the training set: —a norm of the norms of the residuals is calculated; —at least one of optimization, regression analysis or machine-learning is performed to minimize the norm of the norms of the residuals by selecting different settings and iterating the above steps. Settings rendering a minimal training set norm are selected.

A behavior control method for controlling a behavior of a vehicle comprising: specifying a blind-spot region as blind-spot of an environment recognition portion along a travel route for the vehicle; determining a jump-out possibility of a moving object to the travel route from the blind-spot region; performing a possibility reduction behavior to lower the jump-out possibility, in response to that the jump-out possibility is confirmed; and performing a travel behavior compliant with the travel route after starting the possibility reduction behavior.

A device for controlling the movement of a motor vehicle, including a longitudinal controller and a lateral controller which are capable of generating, from first information relating to the road layout and second information relating to the dynamic behaviour of the vehicle, control commands intended for actuators for controlling the longitudinal and lateral movement of the vehicle. The device includes a prediction model which is supplied with the first and second information and is capable of determining future states of the vehicle for future positions of the vehicle over a plurality of iterations defining a future road portion. The model is connected to a module for determining whether driving limit values are violated, which module is capable of determining, for each future state, whether one of the state variables defining the future state reaches or exceeds a driving limit value, and of deducing a future risk situation.

A number of illustrative variations may include a system and method of controlling vehicle slowing while implementing brake-to-steer functionality that may include providing a feed-forward gain on vehicle propulsion torque to achieve or maintain target longitudinal acceleration and replicate the behavior of a vehicle not using brake-to-steer. The system may manipulate propulsion of the vehicle to manage longitudinal acceleration disturbance and speed disturbance during brake-to-steer.

A number of illustrative variations may include a system including brake-to-steer algorithms may achieve lateral control of a vehicle without longitudinal compensation but may also force a vehicle to slow down too rapidly before appropriate lateral movement can be achieved and may deliver an unnatural driving experience for vehicle occupants. A more natural feeling deceleration may be achieved by optimally selecting appropriate transmission shifts to allow for optimal engine speed or electric motor speed and torque based on current vehicle speed thereby reducing undesirably longitudinal disturbance.