LSTM-based lane change prediction using Waymo open motion dataset: The role of vehicle operating space

Lane change prediction is critical for crash avoidance but challenging as it requires the understanding of the instantaneous driving environment.With cutting-edge artificial intelligence and sensing technologies,autonomous vehicles(AVs)are expected to have exceptional perception systems to capture instantaneously their driving environments for predicting lane changes.By exploring the Waymo open motion dataset,this study proposes a framework to explore autonomous driving data and investigate lane change behaviors.In the framework,this study develops a Long Short-Term Memory(LSTM)model to predict lane changing behaviors.The concept of Vehicle Operating Space(VOS)is introduced to quantify a vehicle's instantaneous driving environment as an important indicator used to predict vehicle lane changes.To examine the robustness of the model,a series of sensitivity analysis are conducted by varying the feature selection,prediction horizon,and training data balancing ratios.The test results show that including VOS into modeling can speed up the loss decay in the training process and lead to higher accuracy and recall for predicting lane-change behaviors.This study offers an example along with a methodological framework for transportation researchers to use emerging autonomous driving data to investigate driving behaviors and traffic environments.

Combined Prediction for Vehicle Speed with Fixed Route

Achieving accurate speed prediction provides the most critical support parameter for high-level energy management of plug-in hybrid electric vehicles.Nowadays,people often drive a vehicle on fixed routes in their daily travels and accurate speed predictions of these routes are possible with random prediction and machine learning,but the prediction accuracy still needs to be improved.The prediction accuracy of traditional prediction algorithms is difficult to further improve after reaching a certain accuracy;problems,such as over fitting,occur in the process of improving prediction accuracy.The combined prediction model proposed in this paper can abandon the transitional dependence on a single prediction.By combining the two prediction algorithms,the fusion of prediction performance is achieved,the limit of the single prediction performance is crossed,and the goal of improving vehicle speed prediction performance is achieved.In this paper,an extraction method suitable for fixed route vehicle speed is designed.The application of Markov and back propagation(BP)neural network in predictions is introduced.Three new combined prediction methods,all named Markov and BP Neural Network(MBNN)combined prediction algorithm,are proposed,which make full use of the advantages of Markov and BP neural network algorithms.Finally,the comparison among the prediction methods has been carried out.The results show that the three MBNN models have improved by about 19%,28%,and 29%compared with the Markov prediction model,which has better performance in the single prediction models.Overall,the MBNN combined prediction models can improve the prediction accuracy by 25.3%on average,which provides important support for the possible optimization of plug-in hybrid electric vehicle energy consumption.

Speed prediction models for car and sports utility vehicleat locations along four-lane median divided horizontal curves

Sites with varying geometric features were analyzed to develop the 85 th percentile speed prediction models for car and sports utility vehicle(SUV) at 50 m prior to the point of curvature(PC), PC, midpoint of a curve(MC), point of tangent(PT) and 50 m beyond PT on four-lane median divided rural highways. The car and SUV speed data were combined in the analysis as they were found to be normally distributed and not significantly different. Independent parameters representing geometric features and speed at the preceding section were logically selected in stepwise regression analyses to develop the models. Speeds at various locations were found to be dependent on some combinations of curve length, curvature and speed in the immediately preceding section of the highway. Curve length had a significant effect on the speed at locations 50 m prior to PC, PC and MC. The effect of curvature on speed was observed only at MC. The curve geometry did not have a significant effect on speed from PT onwards. The speed at 50 m prior to PC and curvature is the most significant parameter that affects the speed at PC and MC, respectively. Before entering a horizontal curve, drivers possibly perceive the curve based on its length. Longer curve encourages drivers to maintain higher speed in the preceding tangent section. Further, drivers start experiencing the effect of curvature only after entering the curve and adjust speed accordingly. Practitioners can use these findings in designing consistent horizontal curve for vehicle speed harmony.

Floating Car Data Based Nonparametric Regression Model for Short-Term Travel Speed Prediction

A K-nearest neighbor （K-NN） based nonparametric regression model was proposed to predict travel speed for Beijing expressway. By using the historical traffic data collected from the detectors in Beijing expressways, a specically designed database was developed via the processes including data filtering, wavelet analysis and clustering. The relativity based weighted Euclidean distance was used as the distance metric to identify the K groups of nearest data series. Then, a K-NN nonparametric regression model was built to predict the average travel speeds up to 6 min into the future. Several randomly selected travel speed data series, collected from the floating car data （FCD） system, were used to validate the model. The results indicate that using the FCD, the model can predict average travel speeds with an accuracy of above 90%, and hence is feasible and effective.

GA-LSTM speed prediction-based DDQN energy management for extended-range vehicles

In this paper,a dual deep Q-network(DDQN)energy management model based on long-short memory neural network(LSTM)speed prediction is proposed under the model predictive control(MPC)framework.The initial learning rate and neuron dropout probability of the LSTM speed prediction model are optimized by the genetic algorithm(GA).The prediction results show that the root-mean-square error of the GA-LSTM speed prediction method is smaller than the SVR method in different speed prediction horizons.The predicted demand power,the state of charge(SOC),and the demand power at the current moment are used as the state input of the agent,and the real-time control of the control strategy is realized by the MPC method.The simulation results show that the proposed control strategy reduces the equivalent fuel consumption by 0.0354 kg compared with DDQN,0.8439 kg compared with ECMS,and 0.742 kg compared with the power-following control strategy.The difference between the proposed control strategy and the dynamic planning control strategy is only 0.0048 kg,0.193%,while the SOC of the power battery remains stable.Finally,the hardware-in-the-loop simulation verifies that the proposed control strategy has good real-time performance.

MPC-based path tracking with PID speed control for high-speed autonomous vehicles considering time-optimal travel

In order to track the desired path as fast as possible,a novel autonomous vehicle path tracking based on model predictive control(MPC)and PID speed control was proposed for high-speed automated vehicles considering the constraints of vehicle physical limits,in which a forward-backward integration scheme was introduced to generate a time-optimal speed profile subject to the tire-road friction limit.Moreover,this scheme was further extended along one moving prediction window.In the MPC controller,the prediction model was an 8-degree-of-freedom(DOF)vehicle model,while the plant was a 14-DOF vehicle model.For lateral control,a sequence of optimal wheel steering angles was generated from the MPC controller;for longitudinal control,the total wheel torque was generated from the PID speed controller embedded in the MPC framework.The proposed controller was implemented in MATLAB considering arbitrary curves of continuously varying curvature as the reference trajectory.The simulation test results show that the tracking errors are small for vehicle lateral and longitudinal positions and the tracking performances for trajectory and speed are good using the proposed controller.Additionally,the case of extended implementation in one moving prediction window requires shorter travel time than the case implemented along the entire path.

A Hierarchical LSTM-Based Vehicle Trajectory Prediction Method Considering Interaction Information

Trajectory prediction is an essential component in autonomous driving systems,as it can forecast the future movements of surrounding vehicles,thereby enhancing the decision-making and planning capabilities of autonomous driving systems.Traditional models relying on constant acceleration and constant velocity often experience a reduction in prediction accu-racy as the forecasted timeframe extends.This limitation makes it challenging to meet the demands for medium to long-term trajectory prediction.Conversely,data-driven models,particularly those based on Long Short-Term Memory(LSTM)neural networks,have demonstrated superior performance in medium to long-term trajectory prediction.Therefore,this study introduces a hierarchical LSTM-based method for vehicle trajectory prediction.Considering the difficulty of using a single LSTM model to predict trajectories for all driving intentions,the trajectory prediction task is decomposed into three sequential steps:driving intention prediction,lane change time prediction,and trajectory prediction.Furthermore,given that the driving intent and trajectory of a vehicle are always subject to the influence of the surrounding traffic flow,the predictive model proposed in this paper incorporates the interactional information of neighboring vehicle movements into the model input.The proposed method is trained and validated on the real vehicle trajectory dataset Next Generation Simulation.The results show that the proposed hierarchical LSTM method has a lower prediction error compared to the integral LSTM model.

Spatio-Temporal Wind Speed Prediction Based on Variational Mode Decomposition

Improving short-term wind speed prediction accuracy and stability remains a challenge for wind forecasting researchers.This paper proposes a new variational mode decomposition(VMD)-attention-based spatio-temporal network(VASTN)method that takes advantage of both temporal and spatial correlations of wind speed.First,VASTN is a hybrid wind speed prediction model that combines VMD,squeeze-and-excitation network(SENet),and attention mechanism(AM)-based bidirectional long short-term memory(BiLSTM).VASTN initially employs VMD to decompose the wind speed matrix into a series of intrinsic mode functions(IMF).Then,to extract the spatial features at the bottom of the model,each IMF employs an improved convolutional neural network algorithm based on channel AM,also known as SENet.Second,it combines BiLSTM and AM at the top layer to extract aggregated spatial features and capture temporal dependencies.Finally,VASTN accumulates the predictions of each IMF to obtain the predicted wind speed.This method employs VMD to reduce the randomness and instability of the original data before employing AM to improve prediction accuracy through mapping weight and parameter learning.Experimental results on real-world data demonstrate VASTN’s superiority over previous related algorithms.

电动汽车行驶稳定性模型预测控制仿真研究

为了提高电动汽车行驶的稳定性,提出了一种集成的汽车和车轮稳定性控制系统,并对汽车行驶的稳定性控制效果进行仿真。创建电动汽车模型简图,给出电动汽车动力学方程式。采用模型预测控制技术,设计了一种集成的汽车和车轮稳定性控制器。采用MATLAB软件对电动汽车横摆角速度、行驶速度和侧滑角进行仿真,与无模型预测控制器输出结果进行对比。结果显示:采用无模型预测控制器,电动汽车横摆角速度、行驶速度跟踪误差较大,振动幅度较大,产生的侧滑角较大。采用模型预测控制器,电动汽车横摆角速度、行驶速度跟踪误差较小,振动幅度较小,产生的侧滑角较小。采用模型预测控制器,能够提高电动汽车行驶的稳定性,避免汽车在突然转弯时发生侧翻现象。

多源干扰下的智能车模型预测纵向运动抗干扰控制

智能车纵向运动控制面临模型失配和外部环境变化等多源干扰,影响速度跟踪的精确性,本文针对性提出一种结合扰动观测和模型预测控制(model predict control,MPC)算法的纵向运动抗干扰控制方法。首先,根据车辆纵向动力学模型分析车辆纵向加速度与各项作用力之间的关系,然后将其简化为含多源干扰的质点运动型并设计模型预测控制器作为上层控制器。其次,针对内部未建模动态干扰及外部随机干扰,设计线性扩张状态观测器(linear extended state observe,LESO)进行实时估计,并通过前馈环节实施补偿,且分析了MPC闭环稳定性和LESO收敛性,最终形成扰动补偿和状态反馈的模型预测最优调节控制律。进一步地,为确保控制策略的高效执行,提出1阶自抗扰控制器作为下层控制器,将期望加速度转换为发动机转矩,从而实现对车速的闭环控制。最后,将算法部署在车载微控制单元(microcontroller unit,MCU)上,在多个速度和道路工况下进行实车测试。实验结果表明,所提出的策略可以快速且精确跟踪目标车速,具备良好的抗干扰能力。

基于车速预测的PHEV预测能量管理策略

为解决插电式混合动力汽车预测能量管理策略中车速预测不准确导致车辆燃油经济性降低问题,提出了一种基于麻雀搜索算法优化变分模态分解和长短时神经网络的车速组合预测模型。在模型预测控制架构下采用该预测模型对未来车速进行预测,将全局优化问题转换为预测时域内动力源扭矩优化分配问题,以发动机油耗最小为优化目标,采用动态规划算法对预测时域内的优化问题进行求解。通过仿真表明,所提出的组合预测模型较之于LSTM预测模型预测精度提升了59.57%。同时,基于组合预测模型的预测能量管理策略相较于基于LSTM预测模型的预测控制策略燃油消耗降低了4.58%,相较于基于规则的策略燃油消耗降低了15.1%。

切入场景下基于碰撞风险聚类的改进车速预测方法

切入工况的高精度车速预测是保证自动驾驶切入安全的关键依据。为提高自动驾驶汽车切入工况安全,开展了基于车车耦合风险聚类的切入场景自车速度高精度预测方法的研究。首先,依据实验所得自然驾驶数据进行车辆切入切出片段提取,使用K-means方法依据碰撞风险与加速度关联特征进行聚类分析。其次,基于支持向量机(SVM)模型,对切入切出工况车车交互状态进行在线识别,对切入危险工况进行实时预测。最后,提出基于自回归综合移动平均(ARIMA)模型的改进车速预测方法,结合在线识别结果进行车速在线优化。仿真结果表明,所提出的基于碰撞风险聚类的改进ARIMA车速预测方法对提高切入安全效果明显,较传统的预测方法车辆的碰撞风险降低了10%~20%。研究结果表明,ARIMA模型的改进车速预测方法对提高自动驾驶车切入安全具有重要的研究意义。

Stochastic speed prediction for connected vehicles using improved bayesian networks with back propagation

Advanced vehicular control technologies rely on accurate speed prediction to make ecological and safe decisions. This paper proposes a novel stochastic speed prediction method for connected vehicles by incorporating a Bayesian network(BN) and a Back Propagation(BP) neural network. A BN model is first designed for predicting the stochastic vehicular speed in a priori. To improve the accuracy of the BN-based speed prediction, a BP-based predicted speed error compensation module is constructed by formulating a mapping between the predicted speed and its corresponding prediction error. In the end, a filtering algorithm is developed to smoothen the compensated stochastic vehicular speed. To validate the workings of the proposed approaches in experiments, two typical scenarios are considered: one predecessor vehicle in a double-vehicle scenario and two predecessor vehicles in a multi-vehicle scenario. Simulation results under the considered scenarios demonstrate that the proposed BN-BP fusion method outperforms the BN-based method with respect to the root mean square error, standardized residuals, and R-squared, and the online prediction time of proposed fusion prediction can satisfy a real-time application requirement. The main highlighted contributions of this article are threefold:(1) We put forward an improved BN method, which is combined with a BP neural network, to construct a stochastic vehicular speed prediction method under connected driving;(2) different from existing methods, a unique interconnected framework that consists of a stochastic vehicular speed prediction module, a compensation module, and a speed smoothing module is proposed;(3) extensive simulation studies based on a set of evaluation metrics are illustrated to reveal the advantages and merits of the proposed approaches.

基于DBO-MPC的混合动力汽车能量管理策略

混合动力汽车(hybrid electrical vehicle,HEV)的能量管理策略直接决定了车辆的燃油经济性、驾驶性能和寿命,为解决HEV能量管理策略的最优性与实时行驶工况不确定性之间的矛盾,以混联式HEV为研究对象,提出一种基于模型预测控制(model predictive control,MPC)与蜣螂优化算法(dung beetle optimizer,DBO)的HEV能量管理策略。首先,该策略采用基于堆叠式长短时记忆神经网络(stacked long-short term memory neural network,Stacked LSTM-NN)的车速预测模型预测未来行驶车速。其次,根据预测车速将混合动力汽车的功率分配问题描述为MPC预测范围内的滚动优化问题,提出考虑燃料消耗和电池保护的成本函数,利用DBO算法对预测时域内发动机功率进行优化求解。最后,在城市道路循环(urban dynamometer driving schedule,UDDS)工况下分别对所提策略的车速预测精度和经济性与其他策略进行仿真对比验证。结果表明:与传统LSTM速度预测模型相比,Stacked LSTM速度预测模型的RMSE降低了13.9%,每步平均预测时间减少1 ms;与基于规则的策略相比,基于DBO-MPC的策略模型节油率达到25.3%,同时SOC状态波动更为平稳,对电池的保护效果更好。

考虑前车信息的CNN-BiLSTM的短时车速预测

提出一种考虑跟车信息的基于卷积神经网络(CNN)和双向长短时记忆神经网络(BiLSTM)车速预测模型,引入白鲨优化算法(WSO)对模型的超参数进行优化。综合考虑跟车时的前车信息和其他影响车速的因素,通过驾驶人在环平台采集相关数据,确定了加速踏板开度、制动踏板开度、自车车速、相对车距、相对车速、自车加速度6种变量作为WSO-CNN-BiLSTM模型的输入。通过数据的样本熵值确定变分模态分解的模态个数对数据进行降噪处理。仿真结果显示,考虑前车信息的多输入预测模型相比单一输入预测精度有所提高,且所建立的模型与SVR(support vector regression)、LSTM、CNN和TCN(temporal convolutional network)相比,RMSE值分别降低了63.39%、11.45%、58.45%、42.58%,MAE值分别降低了59.09%、8.09%、57.29%、38.99%,提高了车速预测精度。

Prediction of line-spectrum noise induced by high speed vehicle counter-rotation propellers in water

Line-Spectrum noise of counter-rotation propellers has constructed the main part of the radiated noise of high speed vehicles in water. The line-spectrum noise of the counter-rotation propellers is due to the interaction between fore or aft propeller and wake of the vehicle,and the interaction between fore and aft propeller. Based on a combination of the lifting surface theory and acoustic method, the prediction of line-spectrum noise is presented in this paper.Theoretical calculation method, characteristics and numerical prediction of the line-spectrum noise are detailed too. The effect of different wake and different distance between fore and aft propeller on the propeller noise is also studied by numerical method. The agreement of predicted results compared with existing experimental data is quite satisfactory.

基于中末制导交班点识别的高速高机动飞行器轨迹预测方法

针对高速高机动滑翔飞行器轨迹预测过程中任务场景定义不明确、意图先验信息利用不充分等问题,提出了一种基于中末制导交班点识别的轨迹预测方法。首先,构建高速高机动飞行器滑翔至多个典型中末制导交班点的任务场景,利用准平衡滑翔制导方法生成轨迹数据集。然后,提出了一种基于长短期记忆(Long Short-Term Memory,LSTM)网络的中末制导交班点识别方法,利用跟踪数据构造特征序列,对滑翔轨迹进行初步分类。最后,引入自注意力机制提升序列到序列(Sequence-to-Sequence,Seq2Seq)预测网络的特征提取性能,利用编码-解码的方式对分类后的滑翔轨迹进行长时预测。仿真结果表明,所提出的基于中末制导交班点识别的轨迹预测方法具有较高精度,预测时长为120 s,180 s和240 s时,轨迹误差分别在18.77 km,36.91 km和57.75 km以内;相比于直接利用深度学习模型预测的方法,所提出的预测方法在240 s的预测时长内平均预测误差降低了37.61%,最大预测误差降低了37.34%。

基于元学习方法的道路车速预测模型

准确预测交通道路上的车速,有助于提升交通安全、缓解交通拥挤、促进自动驾驶技术的发展。针对这一需求,本文首次将元学习方法应用到道路车速预测这一领域,研究了基于元学习方法的道路车速预测模型。构建滑动时间窗口,通过6个基础预测器进行初步预测,并用元学习方法给这些预测结果分配一组权值,最终形成组合预测。在公开的数据集上进行了实验,结果表明:该元学习器的预测性能优于目前较为先进的基准模型,为道路车速预测模型提供了更好的解决方案。

Wind speed prediction based on nested shared weight long short-term memory network

With the expansion of wind speed data sets, decreasing model training time is of great significance to the time cost of wind speed prediction. And imperfection of the model evaluation system also affect the wind speed prediction. To address these challenges, a hybrid method based on feature extraction, nested shared weight long short-term memory(NSWLSTM) network and Gaussian process regression(GPR) was proposed. The feature extraction of wind speed promises the best performance of the model. NSWLSTM model reduces the training time of long short-term memory(LSTM) network and improves the prediction accuracy. Besides, it adopted a method combined NSWLSTM with GPR(NSWLSTM-GPR) to provide the probabilistic prediction of wind speed. The probabilistic prediction can provide information that deviates from the predicted value, which is conducive to risk assessment and optimal scheduling. The simulation results show that the proposed method can obtain high-precision point prediction, appropriate prediction interval and reliable probabilistic prediction results with shorter training time on the wind speed prediction.

基于实时交通信息的PHEV模型预测控制策略研究

为提升插电式混合动力汽车(PHEV)的燃油经济性,提出了一种基于实时交通信息的车速预测方法,并以燃油经济性最优为目标,借助动态规划(DP)算法在预测时域内进行实时最优转矩分配,建立基于模型预测控制(MPC)的整车能量管理策略。MATLAB/Simulink仿真平台验证结果表明:与传统车速预测方法相比,基于实时交通信息的车速预测方法的车速预测精确度提高了13.5%;与基于历史车速的模型预测控制策略相比,基于实时交通信息的模型预测控制策略使整车燃油经济性提高了9.5%。