Keywords: Energy management for XEV, Powertrain modeling and control, Exhaust gas after-treatment: catalyst and DPF models, thermal management, SCR control, regeneration control 
Abstract: Combined Energy and Emission Management (CEEM) problems are a class of optimal control problems that aim to minimize operational costs of (hybrid electric) powertrains with after-treatment system subject to constraints on emissions imposed by legislation. In this paper, a parallel-hybrid heavy-duty vehicle with a Variable Turbine Geometry (VTG) and an Exhaust-Gas Recirculation (EGR) system is considered. The CEEM problem is solved using Sequential Quadratic Programming (SQP) for which the powertrain and after-treatment models are approximated as smooth functions. It will be shown that solving the CEEM problem using SQP is computationally much more efficient when compared to other techniques like dynamic programming. It will also be shown that most of the benefits from CEEM come from the hybrid powertrain and not from regulating the VTG and ERG mass flows. Furthermore, zero emission zones and local emission constraints can also be included without too much effort.

Keywords: AI/ML application to automotive and transportation systems
Abstract: Predicting the electrical loads experienced by a battery pack during the 10 hour hotel period of a long haul class 8 mild hybrid truck with a sleeper cab, and using the information to achieve an optimal energy management strategy and controlling the State of Charge (SOC) of battery pack can help in improving it’s freight efficiency. In this work, Machine Learning (ML) based algorithm has been proposed to predict the driver activity during the hotel period. Hence, the power load demanded from the auxiliaries can be predicted. A special kind of Recurrent Neural Network (RNN) called Long and Short Term Memory (LSTM) is used for the prediction task because of its ability to store recurrent information of a small and a large time horizon. To train the LSTM algorithm, the synthetic load profiles are synthesized using rules and observations derived from the existing baseline electrical power load profile of the hotel period. This paper entails the whole process of data synthesis to training the neural network on the synthesized data and the prediction and validation of the power load. The input to the network is a time series of 600 time steps. Dynamic Time Warping (DTW) is used to manipulate the time axis and point wise euclidean distance between the forecast and the test data is used to quantify the accuracy of the model. Then by performing hyper-parameter optimization we find the best

Keywords: AI/ML application to automotive and transportation systems, Energy management for XEV, Battery management systems
Abstract: Energy management in electric vehicles plays a significant role in both reducing energy consumption and limiting the rate of battery capacity degradation. The work summarized in this paper was aimed at exploring of the potential of AI/ML techniques in electrified propulsion control development in designing energy management (EM) controllers. The specific role of the EM strategy was to coordinate delivery of multiple power requests from a modular battery of an Electric Vehicle (EV) to improve range and battery longevity w/o compromising individual load objectives. Within this overall framework, particular application example was integrated EV traction and HVAC controls, where reinforcement learning techniques were adopted within an add-on supervisory scheme to augment existing EV traction and HVAC controls. The proposed supervisory EM controller was structured to monitor current drive parameters and adjust internal HVAC control parameters accordingly to improve energy efficiency and battery SoH w/o affecting driver demand and desired cabin comfort. An empirical battery aging model was incorporated in the problem formulation to address the effect of energy management on long-term battery capacity degradation. Reduced energy consumption and battery aging w/ coordinated controls were demonstrated with several control formulations.

Keywords: Dual fuel control, bio-fuels or bio-gas alternatives, Powertrain modeling and control, Energy management for XEV
Abstract: In this paper, Reactivity Controlled Compression Ignition (RCCI) with an electrically-assisted turbocharger (E-turbo) is investigated. This is a promising concept for future green transport, since it can realize very high thermal efficiencies for a wide range of renewable fuels. The combination of RCCI with an E-turbo requires a new approach to manage the energy flows of the engine due to constraints on the storage of electrical energy. The E-turbo shows most potential to increase the engine's thermal efficiency by improved tracking performance of the desired intake conditions. To exploit the full potential of the E-turbo during transients, a dynamic decoupling feedback controller is designed. A supervisory controller based on Pontryagin's Minimum Principle is composed to maximize the overall engine brake thermal efficiency, while the battery of the E-turbo remains charge sustaining. The supervisory controller determines setpoints for the feedback controller and ensures therefore optimal engine operation during transients. For the simulated, real-world based transient-cycle, fuel savings of 0.64 [%] are realized by the developed supervisory control, while remaining charge sustaining.

Keywords: Energy management for XEV, Vehicle dynamics, control and state estimation, Intelligent transportation systems
Abstract: In this paper, the energy efficiency improvement optimization strategy is explored for large-scale hybrid electric vehicles (HEVs) in a connected environment. Both reducing vehicle speed fluctuation and increasing high efficiency working conditions of HEV powertrain are beneficial for fuel economy improvement. A hierarchical optimization strategy is designed in this paper, where the speed consensus problem is considered in the upper layer and an energy management problem is considered in the lower layer. To deal with optimization of large-scale HEVs, mean field game (MFG) is employed for speed consensus. Meanwhile, model predictive MFG-based control scheme is developed with consideration of distribution predication error caused by the uncertainties of road and traffic. With connection of vehicle to everything (V2X), the real-time distribution can be calculated in the big data center and sent back to individual HEV for model predictive MFG-based controller. Simulations are conducted to show the effectiveness of the proposed strategy.

Keywords: Powertrain modeling and control, Vehicle architecture for XEV, Optimal design and control of XEV
Abstract: Electric vehicles still account for a small share of the total amount of cars on the road. One of the major issues preventing a larger uptake is their higher upfront cost compared to petrol cars. We aim to address this issue by investigating a module-based product-family approach to take full advantage of economy-of-scale strategies, reducing research, development, and production costs of electric vehicles. This paper instantiates a concurrent design optimization framework, whereby different vehicle types share multiple modular powertrain components, whose size is jointly optimized to minimize the overall operational costs instead of being individually tailored. In particular, we focus on sizing battery and electric motors for a family of vehicles equipped with in-wheel motors. First, we identify a convex model of the powertrain, capturing the impact of modules' sizing and multiplicity on the mechanical power demand and the energy consumption of the vehicles. Second, we frame the concurrent powertrain design and operation problem as a second-order conic program that can be efficiently solved with global optimality guarantees. Finally, we showcase our framework for a family of three different vehicles: a city car, a compact car, and an SUV. Our results show that concurrently optimizing shared components increases the operational costs by 3.2% compared to individually tailoring them to each vehicle, a value that could be largely overshadowed by the benefits stemming from using the same components for the entire product family.

Keywords: Optimal design and control of XEV, Modeling and control for electric and electro-magnetic components, Transmissions, brakes, steering, suspension systems
Abstract: This paper instantiates a convex electric powertrain design optimization framework, bridging the gap between high-level powertrain sizing and low-level components design. We focus on the electric motor and transmission of electric vehicles, using a scalable convex motor model based on surrogate modeling techniques. Specifically, we first select relevant motor design variables and evaluate high- fidelity samples according to a predefined sampling plan. Second, using the sample data, we identify a convex model of the motor, which predicts its losses as a function of the operating point and the design parameters. We also identify models of the remaining components of the powertrain, namely a battery and a fixed-gear transmission. Third, we frame the minimum-energy consumption design problem over a drive cycle as a second-order conic program that can be efficiently solved with optimality guarantees. Finally, we showcase our framework in a case study for a compact family car and compute the optimal motor design and transmission ratio. We validate the accuracy of our models with a high-fidelity simulation tool and calculate the drift in battery energy consumption. We show that our model can capture the optimal operating line and the error in battery energy consumption is low. Overall, our framework can provide electric motor design experts with useful starting points for further design optimization.

Keywords: Perception, localization and path planning, Driver-in-the-loop and driver assistance systems, Control, guidance and navigation of autonomous vehicles
Abstract: This paper presents a generic simulation platform with two widely employed software, Simulink and Unreal, to simultaneously simulate the perception in virtual 3D scenarios and system dynamics of automated systems. The proposed CoSim framework improves the accuracy and reduces the development time of automation systems for agricultural crop harvesting and transfer. Strategies using either cameras or LiDAR are supported by the framework. To demonstrate the capability of the proposed CoSim tool, this paper simulates an automated offloading process conducted by a combine-tractor system with a closed-loop controller and a LiDAR-based perception system. The simulation results show that CoSim can be used for both system design and system evaluation.

Keywords: Control, guidance and navigation of autonomous vehicles, Testing and validation, Advanced Driver Assist Systems
Abstract: In this work, we propose a sim2real method to transfer and adapt a nonlinear model predictive controller (NMPC) from simulation to the real target system based on executable digital twin (xDT). The xDT model is a high fidelity vehicle dynamics simulator, executable online in the control parameter randomization and learning process. The parameters are adapted to gradually improve control performance and deal with changing real-world environment. In particular, the performance metric is not required to be differentiable nor analytical with respect to the control parameters and system dynamics are not necessary linearized. Eventually, the proposed sim2real framework leverages altogether online high fidelity simulator, data-driven estimations, and simulation based optimization to transfer and adapt efficiently a controller developed in simulation environment to the real platform. Our experiment demonstrates that a high control performance is achieved without tedious time and labor consuming tuning.

Keywords: Control, guidance and navigation of autonomous vehicles, Perception, localization and path planning, Position, navigation and timing safety and security in automotive systems
Abstract: This paper presents an improved wheel odometry model calibration architecture to increase the accuracy and robustness of the motion estimation of vehicles. Wheel odometry is a robust and cost-effective method, but the accuracy of the estimation is limited by the knowledge of the parameter values. These can be estimated from GNSS and IMU measurements, but the calibration of the nonlinear odometry model in the presence of noise remains an open problem. Due to the nonlinearity, even with Gaussian-type measurement noise on the input wheel speeds, the calibration will be certainly biased. This paper presents an algorithm that takes advantage of the assumption that several measurements are available in a self-driving vehicle, and nowadays the increased computing capacity of computers allows more complex algorithms to be developed. With the proposed architecture, the bias of the model calibration can be reduced significantly through the application of the compensated input signals. The performance of the developed algorithm is demonstrated with detailed validation and test with a real vehicle.