Abstract: A vehicle control method in a hybrid electric vehicle including an internal combustion engine, a battery, an electric motor, and a control unit. The method includes estimating an estimated vehicle velocity trajectory, estimating an initial engine power trajectory, simulating state of charge of the battery with the vehicle velocity trajectory and the initial engine power trajectory, estimating an initial terminal co-state value, simulating backward co-state dynamics using the state of charge and vehicle velocity trajectory, to obtain a resulting co-state trajectory. The co-state trajectory is used to solve a minimization control and propagate state of charge dynamics forward in time. The method includes updating control and the co-state trajectory, adjusting the terminal co-state value, and controlling a usage of the battery and the internal combustion engine. The method can be performed to optimize the engine power trajectory to minimize fuel consumption in real time.

Abstract: A vehicle control method in a hybrid electric vehicle including an internal combustion engine, a battery, an electric motor, and a control unit. The method includes estimating an estimated vehicle velocity trajectory, estimating an initial engine power trajectory, simulating state of charge of the battery with the vehicle velocity trajectory and the initial engine power trajectory, estimating an initial terminal co-state value, simulating backward co-state dynamics using the state of charge and vehicle velocity trajectory, to obtain a resulting co-state trajectory. The co-state trajectory is used to solve a minimization control and propagate state of charge dynamics forward in time. The method includes updating control and the co-state trajectory, adjusting the terminal co-state value, and controlling a usage of the battery and the internal combustion engine. The method can be performed to optimize the engine power trajectory to minimize fuel consumption in real time.

Abstract: The disclosure is directed to solving a full trajectory optimization problem in real-time for a hybrid electric vehicle (HEV) such that future driving conditions and energy usage may be fully considered in determining optimal engine energy usage and battery energy usage in real-time during a trip. An electronic control unit of the HEV may be configured to: receive route information for a route to be driven by the HEV; and after receiving the route information, iterating the operations of: measuring a current state of charge (SOC) of the battery; using at least the measured SOC and an initial co-state value stored in a memory, performing a process to iteratively update the co-state value to obtain an updated co-state value; using at least the updated co-state value, computing an updated control value; and applying the updated control value to control a usage of the battery and the internal combustion engine.

Abstract: A system for control of the air path of an internal combustion engine including a feed-forward controller and a feed-back controller. The feed-forward controller configured to in a sampling period, obtain model parameter values, incorporate the modeled parameter values and reference values into an optimization for a nonlinear model predictive control, perform a Newton method iteration of the optimization in order to determine a solution, and issue commands that control inputs for engine operation based on the solution. The feed-back controller configured to obtain modeled parameter values, obtain measured parameter values based on the operating condition of the engine, incorporate the modeled parameter values, measured parameter values, and reference values into an optimization for a nonlinear model predictive control, perform a Newton method iteration of the optimization in order to determine a solution, and issue commands that control inputs for engine operation based on the solution.

Abstract: Some implementations of the disclosure are directed to reducing or removing time lag in vehicle velocity prediction by training a model for vehicle velocity prediction using labeled features that provide indication of a feature associated with a vehicle acceleration or deacceleration event. In one implementation, a method includes: receiving multiple time series datasets, each of the time series datasets including sensor data, GPS data, and vehicle state data collected over time; extracting features from each of the time series datasets that are indicative of a future velocity of a vehicle; labeling the extracted features of each of the time series datasets to indicate vehicle acceleration or deacceleration events; and after labeling the extracted features of each of the time series datasets, using at least a subset of the extracted and labeled time series datasets to train a machine learning model that predicts vehicle velocity some time into the future.

Abstract: A system including a controller configured to, in each sampling period, minimize a distance of the autonomous vehicle from a target path by solving a constrained control problem, input sensor values and estimators that are calculated based on the sensor values and dynamic models and record the sensor values and the estimators in a memory of the controller, incorporate the sensor values and the estimators into conditions for minimizing the distance of the autonomous vehicle from the target path associated with the constrained control problem, map the conditions for minimizing the distance of the autonomous vehicle from the target path to a non-smooth system using Fischer-Burmeister function, smooth the non-smooth system and apply Newton method iterations to the smoothed system in order to converge on a solution, and issue commands including a steering command that control actuators of the autonomous vehicle based on the solution.

Abstract: Some implementations of the disclosure are directed to reducing or removing time lag in vehicle velocity prediction by training a model for vehicle velocity prediction using labeled features that provide indication of a feature associated with a vehicle acceleration or deacceleration event. In one implementation, a method includes: receiving multiple time series datasets, each of the time series datasets including sensor data, GPS data, and vehicle state data collected over time; extracting features from each of the time series datasets that are indicative of a future velocity of a vehicle; labeling the extracted features of each of the time series datasets to indicate vehicle acceleration or deacceleration events; and after labeling the extracted features of each of the time series datasets, using at least a subset of the extracted and labeled time series datasets to train a machine learning model that predicts vehicle velocity some time into the future.

Abstract: System for controlling an internal combustion engine having a variable geometry turbocharger, an EGR valve and an EGR throttle. The system includes a controller and an engine. In each sampling period, the controller inputs sensor values and estimators that are calculated based on the sensor values and dynamic models and records the sensor values and the estimators in a memory of the controller. The controller incorporates the sensor values and the estimators into conditions for optimality associated with a constrained optimal control problem, maps the conditions for optimality to a non-smooth system using Fischer-Burmeister function, performs Newton method iterations on a smoothed system approximating the non-smooth system in order to converge on a solution, and issues commands that control the EGR valve, the EGR throttle and the variable geometry turbocharger during engine operation.

Abstract: System for controlling an autonomous vehicle having. The system includes a controller. In each sampling period, the controller inputs sensor values and estimators that are calculated based on the sensor values and dynamic models and records the sensor values and the estimators in a memory of the controller. The controller incorporates the sensor values and the estimators into conditions for optimality associated with a constrained optimal control problem, maps the conditions for optimality to a non-smooth system using Fischer-Burmeister function, performs Newton method iterations on a smoothed system approximating the non-smooth system in order to converge on a solution, and issues commands that control actuators during vehicle operation.

Abstract: The disclosure is directed to solving a full trajectory optimization problem in real-time for a hybrid electric vehicle (HEV) such that future driving conditions and energy usage may be fully considered in determining optimal engine energy usage and battery energy usage in real-time during a trip. An electronic control unit of the HEV may be configured to: receive route information for a route to be driven by the HEV; and after receiving the route information, iterating the operations of: measuring a current state of charge (SOC) of the battery; using at least the measured SOC and an initial co-state value stored in a memory, performing a process to iteratively update the co-state value to obtain an updated co-state value; using at least the updated co-state value, computing an updated control value; and applying the updated control value to control a usage of the battery and the internal combustion engine.

Abstract: A system for control of the air path of an internal combustion engine including a feed-forward controller and a feed-back controller. The feed-forward controller configured to in a sampling period, obtain model parameter values, incorporate the modeled parameter values and reference values into an optimization for a nonlinear model predictive control, perform a Newton method iteration of the optimization in order to determine a solution, and issue commands that control inputs for engine operation based on the solution. The feed-back controller configured to obtain modeled parameter values, obtain measured parameter values based on the operating condition of the engine, incorporate the modeled parameter values, measured parameter values, and reference values into an optimization for a nonlinear model predictive control, perform a Newton method iteration of the optimization in order to determine a solution, and issue commands that control inputs for engine operation based on the solution.

Abstract: System for controlling an internal combustion engine having a variable geometry turbocharger, an EGR valve and an EGR throttle. The system includes a controller and an engine. In each sampling period, the controller inputs sensor values and estimators that are calculated based on the sensor values and dynamic models and records the sensor values and the estimators in a memory of the controller. The controller incorporates the sensor values and the estimators into conditions for optimality associated with a constrained optimal control problem, maps the conditions for optimality to a non-smooth system using Fischer-Burmeister function, performs Newton method iterations on a smoothed system approximating the non-smooth system in order to converge on a solution, and issues commands that control the EGR valve, the EGR throttle and the variable geometry turbocharger during engine operation.