Predictive Motion Planning for Intelligent Connected Vehicles Based on Transformer Architecture

doi:10.12141/j.issn.1000-565X.250056

Abstract

Abstract:

Efficient and safe motion planning for intelligent connected vehicles in complex traffic scenarios remains a pivotal challenge in the field of autonomous driving. This research proposed ST-Trans traffic prediction model based on the Transformer architecture and developed a predictive motion planner for intelligent connected vehicles leveraging ST-Trans. The ST-Trans model utilizes the Transformer architecture to mine spatial-temporal evolution patterns from real-time vehicle data and lane segment structural information provided by dynamic high-definition maps, thereby predicting future traffic states of lane segments. It further enhances prediction accuracy by incorporating lane segment connectivity and intersection signal phase information. The model adopts an encoder-decoder framework, where a lane encoder fuses vehicle and lane features, a road encoder models dynamic topological relationships, and a decoder iteratively generates future traffic state sequences. Experimental results demonstrate that ST-Trans outperforms the optimal baseline model by 12.2%, 12.1%, and 3.55 percentage points in terms of mean absolute error(MAE), root mean square error(RMSE), and accuracy, respectively. Based on the predictions from ST-Trans, the proposed predictive motion planner employs a two-layer structure. The lower-layer path planner dynamically selects target points and integrates dynamic programming with quadratic programming to generate smooth paths. The upper-layer speed planner constructs spatio-temporal corridors to compress the solution space and similarly combines dynamic programming and quadratic programming to generate safe efficient, and comfortable speed profiles. This structure significantly reduces the computational complexity of the motion planning task. Simulation experiments were conducted using SUMO and CARLA to evaluate the predictive motion planner. The results indicate that the ST-Trans-based predictive motion planner successfully implements predictive path and speed planning, and outperforms traditional motion planners in terms of safety, efficiency, comfort, and computational speed. The experiments verify that the proposed method effectively shortens the duration of high-risk states, improves traffic throughput and maintains low computational latency.

Key words: dynamic high-definition map, motion planner, traffic prediction model, deep learning

CLC Number:

LI Anran, PAN Yuyan, XU Zhenlin, GAO Bolin, LI Yongxing, YU Hongsheng, CHEN Yanyan. Predictive Motion Planning for Intelligent Connected Vehicles Based on Transformer Architecture[J]. Journal of South China University of Technology(Natural Science Edition), 2026, 54(3): 52-64.

Figures/Tables 10

Fig.1

Fig.2

Fig.3

Table 1

Table 2

Table 3

Fig.4

Fig.5

Fig.6

Table 4

References 39

[1]	GAO B， WAN K， CHEN Q， et al ．A review and outlook on predictive cruise control of vehicles and typical applications under cloud control system［J］．Machine Intelligence Research， 2023， 20（5）： 614-639.
[2]	BAO Z， HOSSAIN S， LANG H， et al ．High-definition map generation technologies for autonomous driving［EB/OL］．（2022-06-24）［2025-11-10］．arXiv preprint，2022：arXiv：.
[3]	DONG J， XU Q， WANG H， et al ．Mixed cloud control testbed：validating vehicle-road-cloud integration via mixed digital twin［J］．IEEE Transactions on Intelligent Vehicles， 2023， 8（4）： 2723-2736.
[4]	冯昶，杜清运，范晓宇，等．高精动态地图基础平台众源更新技术路线研究［J］．测绘地理信息，2023，48（1）： 10-15.
	FENG Chang， DU Qingyun， FAN Xiaoyu， et al ．A crowdsourcing update technology route of HD dynamic map basic platform［J］．Journal of Geomatics，2023，48（1）： 10-15.
[5]	SUNG I， CHOI B， NIELSEN P ．On the training of a neural network for online path planning with offline path planning algorithms［J］．International Journal of Information Management， 2021， 57： 102142/1-9.
[6]	NOTO M， SATO H ．A method for the shortest path search by extended Dijkstra algorithm［C］∥Proceedings of the 2000 IEEE International Conference on Systems，Man and Cybernetics．Nashville：IEEE，2000： 8-11.
[7]	ZHANG Z， ZHAO Z ．A multiple mobile robots path planning algorithm based on A-star and Dijkstra algorithm［J］．International Journal of Smart Home，2014，8（3）：75-86.
[8]	SCHMERLING E， JANSON L， PAVONE M ．Optimal sampling-based motion planning under differential constraints： the drift case with linear affine dynamics［C］∥Proceedings of the 54th IEEE Conference on Decision and Control．Osaka： IEEE， 2015： 5-18.
[9]	KUFFNER J， LAVALLE S ．RRT-connect： an efficient approach to single-query path planning［C］∥Proceedings of the 2 000 IEEE International Conference on Robotics and Automation．San Francisco：IEEE，2 000：995-1001.
[10]	REDDY T， SUPREETHI K ．Optimization of K-means algorithm： ant colony optimization［C］∥Proceedings of the 2017 International Conference on Computing Methodologies and Communication．Erode：IEEE，2017：530-535.
[11]	XIN J， ZHONG J， YANG F， et al ．An improved genetic algorithm for path-planning of unmanned surface vehicle［J］．Sensors， 2019， 19（11）： 2640/1-23.
[12]	PADEN B， ČÁP M， YONG S Z， et al ．A survey of motion planning and control techniques for self-driving urban vehicles［J］．IEEE Transactions on Intelligent Vehicles， 2016， 1（1）： 33-55.
[13]	DUBINS L ．On curves of minimal length with a constraint on average curvature， and with prescribed initial and terminal positions and tangents［J］．American Journal of Mathematics， 1957， 79（3）： 497-516.
[14]	REEDS J， SHEPP L ．Optimal paths for a car that goes both forward and backward［J］．Pacific Journal of Mathematics， 1991， 145（2）： 367-393.
[15]	WANG S， WANG Y， TAN M， et al ．Path planning of underwater biomimetic robot［M］∥Underwater Biomimetic Vehicle-Manipulator System．Singapore：Springer，2023： 89-125.
[16]	LIANG Z， ZHENG G， LI J ．Automatic parking path optimization based on Bezier curve fitting［C］∥Proceedings of the 2012 IEEE International Conference on Automation and Logistics．Zhengzhou：IEEE，2012：583-587.
[17]	ZIEGLER J， BENDER P， DANG T， et al ．Trajectory planning for Bertha—a local， continuous method［C］∥Proceedings of the 2014 IEEE Intelligent Vehicles Symposium．Dearborn：IEEE，2014：450-457.
[18]	WERLING M， KAMMEL S， ZIEGLER J， et al ．Optimal trajectories for time-critical street scenarios using discretized terminal manifolds［J］．The International Journal of Robotics Research， 2012， 31（3）：346-359.
[19]	YU L， SHAO X， WEI Y， et al ．Intelligent land-vehicle model transfer trajectory planning method based on deep reinforcement learning［J］．Sensors，2018，18（9）： 2905/1-22.
[20]	CHAI R， LIU D， LIU T， et al ．Deep learning-based trajectory planning and control for autonomous ground vehicle parking maneuver［J］．IEEE Transactions on Automation Science and Engineering， 2023，20（3）： 1633-1647.
[21]	VASWANI A， SHAZEER N， PARMAR N， et al ．Attention is all you need［C］∥Proceedings of 31st Conference on Neural Information Processing Systems．Long Beach： NIPS， 2017： 6000-6010.
[22]	BARMPOUNAKIS E， GEROLIMINIS N ．On the new era of urban traffic monitoring with massive drone data： the pNEUMA large-scale field experiment［J］．Transportation Research Part C： Emerging Technologies， 2020， 111： 50-71.
[23]	WERLING M， ZIEGLER J， KAMMEL S， et al ．Optimal trajectory generation for dynamic street scenarios in a frenét frame［C］∥Proceedings of the 2010 IEEE International Conference on Robotics and Automation．Anchorage： IEEE， 2010： 987-993.
[24]	MCNAUGHTON M， URMSON C， DOLAN J， et al ．Motion planning for autonomous driving with a conformal spatiotemporal lattice［C］∥Proceedings of the 2011 IEEE International Conference on Robotics and Automation．Shanghai： IEEE， 2011： 4889-4895.
[25]	PIVTORAIKO M， KELLY A ．Differentially constrained motion replanning using state lattices with graduated fidelity［C］∥Proceedings of the 2008 IEEE/RSJ International Conference on Intelligent Robots and Systems．Nice： IEEE， 2008： 2611-2616.
[26]	ZIEGLER J， BENDER P， SCHREIBER M， et al ．Making bertha drive—an autonomous journey on a historic route［J］．IEEE Intelligent Transportation Systems Magazine， 2014， 6（2）： 8-20.
[27]	ZHOU Z， YE L， WANG J， et al ．HiVT： Hierarchical vector transformer for multi-agent motion prediction［C］∥Proceedings of the 2022 IEEE/CVF Confe-rence on Computer Vision and Pattern Recognition．New Orleans： IEEE， 2022： 8813-8823.
[28]	MERCY T， van PARYS R， PIPELEERS G ．Spline-based motion planning for autonomous guided vehicles in a dynamic environment［J］．IEEE Transactions on Control Systems Technology，2017， 26（3）：1-8.
[29]	LI A， XU Z， LI W， et al ．Urban signalized intersection traffic state prediction：a spatial-temporal graph model integrating the cell transmission model and transformer［J］．Applied Sciences，2025，15（5）：2377/1-23.
[30]	HAMED M， AL-MASAEID H， SAID Z ．Short-term prediction of traffic volume in urban arterials［J］．Journal of Transportation Engineering，1995，121（3）：249-254.
[31]	DRUCKER H， BURGES C， KAUFMAN L， et al ．Support vector regression machines［J］．Advances in Neural Information Processing Systems， 1997， 9：779-784.
[32]	ATHEY S， TIBSHIRANI J， WAGER S ．Genera-lized random forests［J］．The Annals of Statistics，2019， 47（2）： 1148-1178.
[33]	YE Q， WONG S， SZETO W ．Short-term traffic speed forecasting based on data recorded at irregular intervals［C］∥Proceedings of 2010 13th International IEEE Annual Conference on Intelligent Transportation Systems. Madeira Island： IEEE，2010： 1541-1546.
[34]	XU Y， LU Y， JI C， et al ．Adaptive graph fusion convolutional recurrent network for traffic forecasting［J］．IEEE Internet of Things Journal，2023，10（13）：11465-11475.
[35]	WU Z， PAN S， LONG G， et al ．Graph wavenet for deep spatial-temporal graph modeling［EB/OL］．（2019-03-31）［2025-11-10］．arXiv preprint， 2019： arXiv：.
[36]	LOPEZ P， BEHRISCH M， BIEKER-WALZ L，et al ．Microscopic traffic simulation using SUMO［C］∥Proceedings of the 2018 21st International Conference on Intelligent Transportation Systems．Maui：IEEE，2018：2575-2582.
[37]	DOSOVITSKIY A， ROS G， CODEVILLA F， et al ．CARLA： an open urban driving simulator［C］∥Proceedings of the 1st Conference on Robot Learning．Mountain View： PMLR， 2017： 1-16.
[38]	FAN H， ZHU F， LIU C， et al ．Baidu Apollo EM motion planner［EB/OL］．（2018-07-20）［2025-11-10］.arXiv preprint，2018：arXiv：.
[39]	WESTHOFEN L， NEUROHR C， KOOPMANN T，et al ．Criticality metrics for automated driving： a review and suitability analysis of the state of the art［J］．Archives of Computational Methods in Engineering，2023，30（1）：1-35.

模型	MAE	RMSE	ACC/%
ARIMA	0.509	1.007	69.91
RF	0.417	0.815	78.95
SVR	0.434	0.849	77.81
FNN	0.337	0.664	84.77
AGCRN	0.316	0.624	86.83
GWN	0.303	0.593	88.43
ST-Trans	0.266	0.521	91.98

模型	MAE	RMSE	ACC/%
ST-Trans_noVT	0.284	0.561	90.02
ST-Trans_noVA	0.274	0.539	90.66
ST-Trans_noLT	0.280	0.563	90.19
ST-Trans_noLD	0.271	0.536	91.00
ST-Trans_noST	0.314	0.627	88.34
ST-Trans	0.266	0.521	91.98

超参数	MAE	RMSE	ACC/%
T_h = 2	0.404	0.791	80.61
T_h = 3	0.353	0.698	85.79
T_h = 4	0.309	0.605	88.84
T_h = 5	0.286	0.566	90.29
T_h = 6	0.272	0.538	91.16
T_h = 7	0.266	0.521	91.98
T_h = 8	0.264	0.517	92.25
T_p = 10	0.248	0.497	92.96
T_p = 15	0.254	0.508	92.50
T_p = 20	0.266	0.521	91.98
T_p = 25	0.289	0.568	89.98
d_h = 16	0.274	0.537	90.81
d_h = 32	0.269	0.529	91.62
d_h = 64	0.266	0.521	91.98
d_h = 128	0.264	0.518	92.11
N_h = 2	0.279	0.560	90.34
N_h = 4	0.268	0.528	91.69
N_h = 8	0.266	0.521	91.98
N_h = 16	0.268	0.526	91.77

场景	行为	模型	碰撞时间			平均通行时长	平均计算延误
场景	行为	模型	高风险状态	风险预警状态	安全状态	平均通行时长	平均计算延误
低密度	直行	EM规划器	1.76	3.27	6.12	11.15	0.087
	直行	ST-Trans规划器	0.65	6.31	3.07	10.03	0.069
	左转	EM规划器	2.55	3.13	10.95	16.63	0.102
	左转	ST-Trans规划器	0.98	9.65	4.53	15.16	0.075
	右转	EM规划器	2.29	3.57	7.60	13.46	0.095
	右转	ST-Trans规划器	1.14	7.34	3.59	12.07	0.073
高密度	直行	EM规划器	2.52	4.33	8.42	15.27	0.094
	直行	ST-Trans规划器	1.04	8.79	4.23	14.06	0.070
	左转	EM规划器	3.95	4.53	14.86	23.34	0.106
	左转	ST-Trans规划器	1.98	13.14	6.23	21.35	0.077
	右转	EM规划器	3.75	4.91	10.46	19.12	0.102
	右转	ST-Trans规划器	1.92	10.09	5.53	17.54	0.076