Senqi Tan

Personalized policy imitation of autonomous vehicles (AVs) is an important technique to lessen human–machine conflict and improve AVs’ predictability to other traffic participants. However, existing imitation learning methods, both model-free and model-based, fail to solve this issue effectively and efficiently if only one demonstration of the target driver is available. We propose a personalized lane change policy for autonomous driving, which can end-to-end learn human drivers’ behavior from a few demonstrations. To this end, combining model-free imitation learning and a model-based differentiable optimization method, we establish a policy generator with learnable personalized parameters in the cost function. By differentiating the equations that satisfy Pontryagin’s Maximum Principle (PMP), we solve the gradient of imitation loss with respect to learnable personalized parameters. The gradient is used to update and ultimately recover these personalized parameters in an end-to-end learning fashion. Here, “end-to-end” means that the personalized parameters are recovered by gradient-based optimization inside a structured model-based control framework, rather than by learning direct observation-to-action mapping. With these recovered parameters, the proposed policy generator imitates drivers’ behavior. The results prove the following: (1) the proposed method can mimic different drivers’ lane change policies with both sufficient data and only one demonstration by recovering the personalized cost function automatically, and (2) by recovering the personalized cost function automatically, the proposed method achieves a comparable final imitation loss to the differentiable-MPC baseline, with faster and more stable training convergence and a substantially lower computational cost in the one-demonstration setting.

The distributed drive electric platform is extensively utilized in special-purpose unmanned ground vehicles due to exceptional maneuverability and obstacle navigation capabilities in off-road environments. This paper comes up with an energy-saving torque allocation strategy to enhance path-tracking precision and energy utilization efficiency. For mitigating the conflict between path tracking and energy management in torque distribution, a hierarchical control framework is proposed for total demanding torque control, inter-axle torque distribution, and inter-wheel torque allocation. The inter-axle torque distribution is determined by considering the total motors efficiency region reflection to enhance energy conversion efficiency. This approach ensures that torque is allocated to maximize energy conversion efficiency. Besides, the inter-wheel torque allocation is optimized using quadratic programming, considering factors such as tire vertical load distribution and path tracking error. By decoupling these control aspects, the proposed framework ensures optimal performance in both path tracking and energy management. In contrast, a traditional torque allocation method is also presented. Finally, the simulation experiments are conducted in MATLAB/Simulink and TruckSim with specialized off-road working cycles. The numerical experiment results demonstrate that the presented strategy considerably ensures an optimal torque allocation for path tracking while simultaneously enhances the overall efficiency of the distributed drive system compared with the conventional approach.

The controller design of vehicle systems depends on accurate reference index input. Considering information fusion and feature extraction based on existing data settings in the time domain, if reasonable input is selected for prediction to obtain accurate information of future state, it is of great significance for control decision-making, system response, and driver’s active intervention. In this paper, the nonlinear dynamic model of the four-wheel steering vehicle system was built, and the Long Short-Term Memory (LSTM) network architecture was established. On this basis, according to the real-time data under different working conditions, the information correction calculation of variable time-domain length was carried out to obtain the real-time state input length. At the same time, the historical state data of coupled road information was adopted to train the LSTM network offline, and the acquired real-time data state satisfying the accuracy was used as the LSTM network input to carry out online prediction of future confidence information. In order to solve the problem of mixed sensitivity of the system, a robust controller for vehicle active steering was designed with the sideslip angle of the centroid of 0, and the predicted results were used as reference inputs for corresponding numerical calculation verification. Finally, according to the calculated results, the robust controller with information prediction can realize the system stability control under coupling conditions on the premise of knowing the vehicle state information in advance, which provides an effective reference for controller response and driver active manipulation.

Unstructured off-road environments with complex terrain obstacles and pavement properties bring obvious challenges for special purpose autonomous vehicle control. A cascade direct yaw moment control strategy (CDYC), which contains a main loop and a servo loop, is proposed to enhance the accuracy and stability of an independent eight in-wheel motor-driven autonomous vehicle with rear-wheel steering (8WD/RWS). In the main loop, double PID controllers are designed to generate the desired drive moment and yaw rate. In the servo loop, the quadratic programming (QP) algorithm with the tire force boundaries optimally allocates the demanded yaw moment to individual wheel torques. The 8WD/RWS prototype is virtually established using TruckSim and serves as the control object for co-simulation. The proposed cascade controller is verified by simulations in customized off-road driving scenarios. The simulation results show that the proposed control architecture can effectively enhance the path-tracking ability and handling stability of the 8WD/RWS, as to ensure the maneuverability and control stability under extreme off-road conditions.

In improving the efficiency of powertrain systems and ride comfort for electric vehicles (EVs), the transmission model is required to enable more accessible and more straightforward control of such vehicles. In this study, a wet-type, two-speed, dual...

Electric dump trucks (EDTs) are expected to work in fast-changing conditions of torque and speed, which causes the inefficient working status of traction motors. To improve the economic performance of the dump truck, a novel integrated dual-motor transmission (IDMT) is proposed in this article. The energy consumption level of this kind of system relies on a wise energy management strategy. This article presents a dynamic power distribution strategy (DPDS) for the proposed system. By coordinating the output torque allocation of the two motors based on the efficiency map partition and optimal dynamic torque allocation ratio, the proposed strategy is more effective for real-time application compared to the previously proposed enumeration strategy. In addition, a particle swarm optimization is carried out to search for the optimal combination of the gear ratios for the selected EDT. The dynamic model of the IDMT and a detailed mathematical model of the EDT are developed to validate the performance of the proposed design and strategy. The simulation and experiment results show the energy-save and real-time ability of the developed strategy and indicate that the proposed structure can appreciably improve economic performance and enhance battery life.

This paper presents a control method based the lateral interconnected air suspension system, in order to improve the road handling of vehicles. A seven-DOF (Degree of freedom) full-vehicle model has been developed, which considers the features of the interconnected air suspension system, for example, the modeling of the interconnected pipelines and valves by considering the throttling and hysteresis effects. On the basis of the well-developed model, a sliding mode controller has been designed, with a focus on constraining and minimizing the roll motion of the sprung mass caused by the road excitations or lateral acceleration of the vehicle. Moreover, reasonable road excitations have been generated for the simulation based on the coherence of right and left parts of the road. Afterwards, different simulations have been done by applying both bumpy and random road excitations with different levels of roughness and varying vehicle lateral accelerations. The simulation results indicate that the interconnected air suspension without control can improve the ride comfort, but worsen the road handling performance in many cases. However, by applying the proposed sliding mode controller, the road handling of the sprung mass can be improved by 20% to 85% compared with the interconnected or non-interconnected mode at a little cost of comfort.

Research demonstrated that the application of a multi-speed transmission could improve the dynamic and economic performance of electric vehicles. This paper deals with a novel multi-speed transmission for the electric dump truck (EDT) operating on the structure route (SR), which has a definite starting point and end point without complex traffic conditions. To optimize the gear ratio and shift schedule to reduce energy consumption in such conditions, the mathematical model of the transmission and the dynamic model of the EDT are initially required. Following this, the shift schedule is presented according to the motor efficiency map. After that, the gear ratio optimization is carried out by a particle swarm optimization (PSO) algorithm. Moreover, the proposed EDT is compared with an EDT with a single-speed transmission. The simulation results show that the energy consumption is reduced by 6.1%.