识别交通模式是出行即服务一体化出行中重要的任务，针对现有交通模式识别算法精度有待提高以及实时性应用需求方面的挑战，提出一种基于自适应池增强注意力机制的交通模式实时识别算法。该算法基于点云网络，通过引入因果卷积和因果池化高效学习当前及历史信息，从而实现交通模式实时识别。通过在模型框架中嵌入自适应池增强注意力模块，计算不同特征之间的权重图，进一步提升特征建模能力。同时，该算法还融合了运动特征与地理特征，有效提高了对公交车与小汽车交通模式的识别精度。实验结果表明，该算法在精度方面表现优异，相较其他一阶段方法，其识别精度提升约0.05；与最新的FPbiLSTM等两阶段模型相比，参数量仅为0.167，更加轻量化，更适合部署于移动端设备。
Identifying travel patterns is a crucial task in integrated mobility as a service. To address the limitations of existing traffic travel pattern recognition algorithms, namely insufficient accuracy and real-time application demands, this paper proposed a real-time travel pattern recognition algorithm based on a self adaptive pooling enhanced attention mechanism. Based on the point cloud network, the proposed algorithm efficiently learned current and historical information by incorporating causal convolution and causal pooling, thereby achieving real-time travel pattern recognition. A self adaptive pooling enhanced attention module was further embedded into the framework model to calculate the weight map among different features, thereby enhancing the feature modeling capability. Additionally, the algorithm integrated both motion and geographical features, which effectively improved the recognition accuracy of bus and car travel patterns. Experimental results show that the proposed algorithm achieves superior accuracy. Compared with other one-stage methods, its recognition accuracy is improved by approximately 0.05. Compared with the latest two-stage models such as FPbiLSTM, the parameter count of the proposed algorithm is only 0.167 that of these models, making the proposed approach more lightweight and suitable for deployment on mobile devices.
[1] Li J, Pei X, Wang X, et al. Transportation mode identification with GPS trajectory data and GIS information [J]. Tsinghua Science and Technology, 2021, 26(4): 403-416.
[2] Cheng X, Zhang R, Zhou J, et al. Deeptransport: learning spatial-temporal dependency for traffic condition forecasting [C]//International Joint Conference on Neural Networks (IJCNN), 2018: 1-8.
[3] Sun F, Hao W, Zou A, et al. A survey on spatio-temporal series prediction with deep learning: taxonomy, applications, and future directions [J]. Neural Computing and Applications, 2024, 36(17): 9919-9943.
[4] Dabbas H, Friedrich B. Benchmarking machine learning algorithms by inferring transportation modes from unlabeled GPS data [J]. Transportation Research Procedia, 2022, 62: 383-392.
[5] Zheng Y, Li Q, Chen Y, et al. Understanding mobility based on GPS data [C]//10th International Conference on Ubiquitous Computing, 2008: 312-321.
[6] Xiao G, Juan Z, Zhang C. Detecting trip purposes from smartphone-based travel surveys with artificial neural networks and particle swarm optimization [J]. Transportation Research Part C: Emerging Technologies, 2016, 71: 447-463.
[7] Dabiri S, Heaslip K. Inferring transportation modes from GPS trajectories using a convolutional neural network [J]. Transportation Research Part C: Emerging Technologies, 2018, 86: 360-371.
[8] James J. Travel mode identification with GPS trajectories using wavelet transform and deep learning [J]. IEEE Transactions on Intelligent Transportation Systems, 2020, 22(2): 1093-1103.
[9] Tang Q, Jahan K, Roth M. Deep CNN-BiLSTM model for transportation mode detection using smartphone accelerometer and magnetometer [C]//IEEE Intelligent Vehicles Symposium (Ⅳ), 2022: 772-778.
[10] Ma Y, Guan X, Cao J, et al. A multi-stage fusion network for transportation mode identification with varied scale representation of GPS trajectories [J]. Transportation Research Part C: Emerging Technologies, 2023, 150: 104088.
[11] Tang Q, Cheng H. Feature pyramid BiLSTM: using smartphone sensors for transportation mode detection [J]. Transportation Research Interdisciplinary Perspectives, 2024, 26: 101181.
[12] Moiseeva A, Jessurun J, Timmermans H. Semiautomatic imputation of activity travel diaries: use of global positioning system traces, prompted recall, and context-sensitive learning algorithms [J]. Transportation Research Record, 2010, 2183(1): 60-68.
[13] Prelipcean A C, Gidófalvi G, Susilo Y O. Mobility collector [J]. Journal of Location Based Services, 2014, 8(4): 229-255.
[14] Jiang X, De Souza E N, Pesaranghader A, et al. TrajectoryNet: an embedded GPS trajectory representation for point-based classification using recurrent neural networks [C]// 27th Annual International Conference on Computer Science and Software Engineering (CASCON’17), 2017: 1-9.
[15] Dutta S, Patra B K. Inferencing transportation mode using unsupervised deep learning approach exploiting GPS point-level characteristics [J]. Applied Intelligence, 2023, 53(10): 12489- 12503.
[16] Li R, Yang Z, Pei X, et al. A novel one-stage approach for pointwise transportation mode identification inspired by point cloud processing [J]. Transportation Research Part C: Emerging Technologies, 2023, 152: 104127.
[17] Mücke N T. Deep learning for real-time inverse problems and data assimilation with uncertainty quantification for digital twins [D]. Utrecht: Utrecht University, 2024.
[18] Muhasshanah M, Tohir M, Ningsih D A, et al. Comparison of the performance results of c4.5 and random forest algorithm in data mining to predict childbirth process [J]. CommIT (Communication and Information Technology) Journal, 2023, 17(1): 51-59.
[19] Feng T, Timmermans H J. Transportation mode recognition using GPS and accelerometer data [J]. Transportation Research Part C: Emerging Technologies, 2013, 37: 118-130.
[20] Roy S, Singh Y P, Biswas U, et al. Machine learning in smart transportation systems for mode detection [C]//2021 IEEE 18th India Council International Conference (INDICON), 2021: 1-6.
[21] Zheng Y, Xie X, Ma W Y. GeoLife: a collaborative social networking service among user, location and trajectory [J]. IEEE Data Eng Bull, 2010, 33(2): 32-39.
