Shiying Dong
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Shiying Dong has a total of 37 co-authors. Here are the top co-authors by number of shared publications:
- Bingzhao Gao
Tongji University
14 shared publications - Hong Chen
Tongji University
11 shared publications - Qifang Liu
Department of Control Science and Engineering, Jilin University, Changchun, China
6 shared publications - Jinlong Hong
School of Automotive Studies, Tongji University, Shanghai 201804, China
4 shared publications - Lulu Guo
Department of Control Science and Engineering, Tongji University, Shanghai, China
4 shared publications - Hongqing Chu
School of Automotive Studies, Tongji University, Shanghai, China
3 shared publications
Latest Publications
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Control Systems Technology
- Vol. 34 (5) , 2925-2932
A Fast Semidefinite Convex Relaxation for Optimal Control Problems With Spatio-Temporal Constraints
- 1 September 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Control Systems Technology
- Vol. 34 (5) , 2925-2932
A Fast Semidefinite Convex Relaxation for Optimal Control Problems With Spatio-Temporal Constraints
- 8 January 2026
- Published by Springer Nature in Automotive Innovation
- Vol. 9 (1) , 65-77
Fast Convex Optimization of Velocity Planning and Powertrain Control for Eco-Driving
- 8 January 2026
- Published by Springer Nature in Automotive Innovation
- Vol. 9 (1) , 65-77
Fast Convex Optimization of Velocity Planning and Powertrain Control for Eco-Driving
- 24 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 73 (5) , 7619-7630
Physics-Informed LSTM Network for Integrated State Estimation and Motion Control of Autonomous Vehicles
In autonomous vehicles (AVs), optimization-based coordinated control of multiple actuators entails a substantial computational burden. Moreover, the complexity of driving scenarios makes the design of observers for unmeasurable states particularly challenging. To overcome these limitations, we introduce a physics-informed long short-term memory (PI-LSTM) network that learns a mapping from historical onboard sensor data to control actions within a physics-guided framework. Specifically, the PI-LSTM forms an integrated architecture that unifies state estimation and motion control. The upstream LSTM estimator infers unobservable vehicle states from measurable sensor signals, and its outputs are explicitly propagated to the downstream LSTM-based model predictive control (MPC) solver, which is trained within a physics-informed framework that incorporates discrete-time vehicle dynamics, constraints, and cost functions to compute optimal control actions over a finite prediction horizon. By embedding physics-based priors and preserving explicit signal pathways, the interpretability, robustness, and generalization of PI-LSTM in safety-critical applications such as chassis control are enhanced. Extensive co-simulations and real vehicle experiments demonstrate that the proposed method improves closed-loop control performance by more than 5% compared with baseline methods, while achieving real-time inference at 100 Hz on embedded vehicle controllers. This work highlights the potential of hybrid data- and physics-driven approaches for advancing state estimation and motion control in AVs.
- 24 December 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Industrial Electronics
- Vol. 73 (5) , 7619-7630
Physics-Informed LSTM Network for Integrated State Estimation and Motion Control of Autonomous Vehicles
In autonomous vehicles (AVs), optimization-based coordinated control of multiple actuators entails a substantial computational burden. Moreover, the complexity of driving scenarios makes the design of observers for unmeasurable states particularly challenging. To overcome these limitations, we introduce a physics-informed long short-term memory (PI-LSTM) network that learns a mapping from historical onboard sensor data to control actions within a physics-guided framework. Specifically, the PI-LSTM forms an integrated architecture that unifies state estimation and motion control. The upstream LSTM estimator infers unobservable vehicle states from measurable sensor signals, and its outputs are explicitly propagated to the downstream LSTM-based model predictive control (MPC) solver, which is trained within a physics-informed framework that incorporates discrete-time vehicle dynamics, constraints, and cost functions to compute optimal control actions over a finite prediction horizon. By embedding physics-based priors and preserving explicit signal pathways, the interpretability, robustness, and generalization of PI-LSTM in safety-critical applications such as chassis control are enhanced. Extensive co-simulations and real vehicle experiments demonstrate that the proposed method improves closed-loop control performance by more than 5% compared with baseline methods, while achieving real-time inference at 100 Hz on embedded vehicle controllers. This work highlights the potential of hybrid data- and physics-driven approaches for advancing state estimation and motion control in AVs.
- 31 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 26 (11) , 18581-18596
Deep Learning-Based Mixed-Integer Co-Optimization of Velocity Planning and Powertrain Control in Urban Environment
This paper proposes an online co-optimization strategy for velocity planning and powertrain control in autonomous vehicles (AVs), aiming to improve energy efficiency and travel time in urban driving. A hierarchical eco-driving framework is designed to incorporate both long-horizon traffic signal information and short-horizon traffic dynamics. In the upper layer, velocity planning is formulated as a mixed-integer optimal control problem (MIOCP) in the distance domain. Binary variables are introduced using the big-M method to model disjunctive constraints associated with signalized intersection crossing decisions. The lower layer ensures velocity eco-tracking and car-following safety in the time domain while co-optimizing gearshift, traction, and braking torque, also resulting in an MIOCP. To address the real-time computation challenge, a two-stage optimization method is introduced. First, a neural network (NN) is trained offline via supervised learning to predict feasible integer strategies based on MIOCP parameters. Then, the predicted integers are fixed online, and the resulting problem is solved as a nonlinear programming (NLP) problem. The effectiveness of the proposed method is thoroughly validated through extensive simulation studies under various urban scenarios, including multiple signalized intersections, varying road speed limits, and diverse preceding vehicle (PV) behaviors. The results demonstrate that the proposed approach enables safe and adaptive eco-driving, obtaining a 5.97% average improvement in energy efficiency over the intelligent driver model, while accelerating computation speed by 1–2 orders of magnitude compared to Bonmin and achieving millisecond-level computational time.
- 31 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Intelligent Transportation Systems
- Vol. 26 (11) , 18581-18596
Deep Learning-Based Mixed-Integer Co-Optimization of Velocity Planning and Powertrain Control in Urban Environment
This paper proposes an online co-optimization strategy for velocity planning and powertrain control in autonomous vehicles (AVs), aiming to improve energy efficiency and travel time in urban driving. A hierarchical eco-driving framework is designed to incorporate both long-horizon traffic signal information and short-horizon traffic dynamics. In the upper layer, velocity planning is formulated as a mixed-integer optimal control problem (MIOCP) in the distance domain. Binary variables are introduced using the big-M method to model disjunctive constraints associated with signalized intersection crossing decisions. The lower layer ensures velocity eco-tracking and car-following safety in the time domain while co-optimizing gearshift, traction, and braking torque, also resulting in an MIOCP. To address the real-time computation challenge, a two-stage optimization method is introduced. First, a neural network (NN) is trained offline via supervised learning to predict feasible integer strategies based on MIOCP parameters. Then, the predicted integers are fixed online, and the resulting problem is solved as a nonlinear programming (NLP) problem. The effectiveness of the proposed method is thoroughly validated through extensive simulation studies under various urban scenarios, including multiple signalized intersections, varying road speed limits, and diverse preceding vehicle (PV) behaviors. The results demonstrate that the proposed approach enables safe and adaptive eco-driving, obtaining a 5.97% average improvement in energy efficiency over the intelligent driver model, while accelerating computation speed by 1–2 orders of magnitude compared to Bonmin and achieving millisecond-level computational time.
- 22 July 2025
- Published by Elsevier in Control Engineering Practice
- Vol. 164
Computationally efficient predictive cruise control of electric vehicles with nonsmooth system
- 22 July 2025
- Published by Elsevier in Control Engineering Practice
- Vol. 164
Computationally efficient predictive cruise control of electric vehicles with nonsmooth system
- 10 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 74 (12) , 18819-18832
Deep Learning-Based Real-Time Model Predictive Motion Control for Distributed Drive Electric Vehicles Under Extreme Conditions
The motion control of distributed drive electric vehicles (DDEVs) under extreme operating conditions is significantly affected by the strong nonlinearities in vehicle dynamics. A novel hierarchical control architecture is proposed to enhance the control performance under such conditions. The upper-level deep learning-based model predictive motion controller determines the front wheel steering angle, longitudinal force, and additional yaw moment based on the vehicle's current state, target trajectory, and sideslip angle at the center of gravity (CG) constraints. It integrates low-fidelity, interpretable physical models with neural networks (NNs), leveraging automatic differentiation to backpropagate the nonlinear model predictive control (NMPC) loss function through a differentiable nonlinear dynamics model to obtain policy gradients, while using tailored NN output layers and an augmented Lagrangian method to ensure the satisfaction of control and state constraints. The lower-level torque distributor is constructed to prioritize vehicle stability by minimizing tire adhesion utilization. High-fidelity co-simulations and hardware-in-the-loop (HIL) experiments including an embedded onboard controller demonstrate the effectiveness of the proposed hierarchical control scheme in terms of real-time computation and closed-loop control performance. The real-vehicle test results also demonstrate that our algorithm can effectively handle state constraints and ensure vehicle stability under extreme conditions.
- 10 July 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Vehicular Technology
- Vol. 74 (12) , 18819-18832
Deep Learning-Based Real-Time Model Predictive Motion Control for Distributed Drive Electric Vehicles Under Extreme Conditions
The motion control of distributed drive electric vehicles (DDEVs) under extreme operating conditions is significantly affected by the strong nonlinearities in vehicle dynamics. A novel hierarchical control architecture is proposed to enhance the control performance under such conditions. The upper-level deep learning-based model predictive motion controller determines the front wheel steering angle, longitudinal force, and additional yaw moment based on the vehicle's current state, target trajectory, and sideslip angle at the center of gravity (CG) constraints. It integrates low-fidelity, interpretable physical models with neural networks (NNs), leveraging automatic differentiation to backpropagate the nonlinear model predictive control (NMPC) loss function through a differentiable nonlinear dynamics model to obtain policy gradients, while using tailored NN output layers and an augmented Lagrangian method to ensure the satisfaction of control and state constraints. The lower-level torque distributor is constructed to prioritize vehicle stability by minimizing tire adhesion utilization. High-fidelity co-simulations and hardware-in-the-loop (HIL) experiments including an embedded onboard controller demonstrate the effectiveness of the proposed hierarchical control scheme in terms of real-time computation and closed-loop control performance. The real-vehicle test results also demonstrate that our algorithm can effectively handle state constraints and ensure vehicle stability under extreme conditions.
- 23 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Transportation Electrification
- Vol. 11 (4) , 9532-9543
Mixed-Integer Optimal Control With Disjunctive Characteristic Time Constraints for Energy-Optimal Speed Planning
This article considers the energy optimization problem with multiple time window constraints, which is motivated by the eco-driving of connected and automated vehicles (CAVs) in urban traffic. Timely crossing of signalized intersections defines complex spatio-temporal constraints, and multiple available open time windows introduce the discontinuous feasible domains of crossing times. The CAVs should select an open time window for each intersection, determine the crossing times, and optimize the speed profiles accordingly, which turns out to be a hard-to-solve optimization problem. To address this challenge, this article proposes a unifying approach that transforms the energy optimization problem into a novel mixed-integer optimal control problem with characteristic time (MIOCP-CT) constraints. This approach facilitates the simultaneous optimization of open time windows, crossing times, and speed profiles within a single optimization problem, which can be solved by our proposed novel numerical solution procedure. The application of the proposed method in the eco-driving of CAVs approaching multiple signalized intersections demonstrates its performance within real-world urban driving cycles. The experimental results of the hardware-in-the-loop (HiL) demonstrate a superior energy-saving performance of 5.83% compared to existing methods, while also achieving real-time feasibility within 100 ms.
- 23 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Transportation Electrification
- Vol. 11 (4) , 9532-9543
Mixed-Integer Optimal Control With Disjunctive Characteristic Time Constraints for Energy-Optimal Speed Planning
This article considers the energy optimization problem with multiple time window constraints, which is motivated by the eco-driving of connected and automated vehicles (CAVs) in urban traffic. Timely crossing of signalized intersections defines complex spatio-temporal constraints, and multiple available open time windows introduce the discontinuous feasible domains of crossing times. The CAVs should select an open time window for each intersection, determine the crossing times, and optimize the speed profiles accordingly, which turns out to be a hard-to-solve optimization problem. To address this challenge, this article proposes a unifying approach that transforms the energy optimization problem into a novel mixed-integer optimal control problem with characteristic time (MIOCP-CT) constraints. This approach facilitates the simultaneous optimization of open time windows, crossing times, and speed profiles within a single optimization problem, which can be solved by our proposed novel numerical solution procedure. The application of the proposed method in the eco-driving of CAVs approaching multiple signalized intersections demonstrates its performance within real-world urban driving cycles. The experimental results of the hardware-in-the-loop (HiL) demonstrate a superior energy-saving performance of 5.83% compared to existing methods, while also achieving real-time feasibility within 100 ms.
- Open Access
- 1 January 2025
- Published by Institution of Engineering and Technology (IET) in IET Intelligent Transport Systems
- Vol. 19 (1)
Lane‐Changing‐Enabled Eco Approach Control for Autonomous Vehicles at a Signalized Intersection in Mixed Traffic Environment
In complex urban traffic environments, vehicle‐to‐everything technology can reduce energy consumption and optimize travel time for vehicles at a signalized intersection. We propose a lane‐changing‐enabled eco approach control strategy that considers constraints from surrounding vehicles, traffic lights, and queues ahead. Our strategy adopts a hierarchical receding horizon control framework that provides the connected and autonomous vehicle with optimal lane and speed planning for adapting to dynamic traffic environment. In the upper level, we set up virtual traffic light for each lane via a mixed vehicle platoon model, utilizing their signal phase time constraints and constraints of surrounding vehicles to formulate a mixed‐integer nonlinear programming problem to obtain optimal lane and speed reference. In the lower level, a car‐following model is conducted in time domain and the reference speed is put into the controller to obtain the optimal velocity. Numerical experiment results show that our strategy is superior to regular eco‐approach and departure strategy and lane‐changing‐enabled car‐following strategy in terms of energy consumption reduction in the traffic scenarios without lane changing. In the traffic scenarios with lane changing, our strategy improves traffic efficiency by reducing travel time.
- 1 January 2025
- Published by Institution of Engineering and Technology (IET) in IET Intelligent Transport Systems
- Vol. 19 (1)
Open AccessLane‐Changing‐Enabled Eco Approach Control for Autonomous Vehicles at a Signalized Intersection in Mixed Traffic Environment
In complex urban traffic environments, vehicle‐to‐everything technology can reduce energy consumption and optimize travel time for vehicles at a signalized intersection. We propose a lane‐changing‐enabled eco approach control strategy that considers constraints from surrounding vehicles, traffic lights, and queues ahead. Our strategy adopts a hierarchical receding horizon control framework that provides the connected and autonomous vehicle with optimal lane and speed planning for adapting to dynamic traffic environment. In the upper level, we set up virtual traffic light for each lane via a mixed vehicle platoon model, utilizing their signal phase time constraints and constraints of surrounding vehicles to formulate a mixed‐integer nonlinear programming problem to obtain optimal lane and speed reference. In the lower level, a car‐following model is conducted in time domain and the reference speed is put into the controller to obtain the optimal velocity. Numerical experiment results show that our strategy is superior to regular eco‐approach and departure strategy and lane‐changing‐enabled car‐following strategy in terms of energy consumption reduction in the traffic scenarios without lane changing. In the traffic scenarios with lane changing, our strategy improves traffic efficiency by reducing travel time.