پژوهشنامه حمل و نقل

پژوهشنامه حمل و نقل

بررسی عملکردِ انواع شبکۀ عصبی بازگشتی در پیش‌بینیِ داده‌های سریِ زمانی در حمل‌‌ و‌ نقل؛ نوع داده: مسیر حرکت عابرپیاده در پیاده‌رو

نوع مقاله : مقاله پژوهشی

نویسندگان
1 دانشیار، دانشکده مهندسی عمران، دانشگاه صنعتی خواجه نصیرالدین طوسی، تهران، ایران
2 دانشجوی کارشناسی ارشد، دانشگاه صنعتی خواجه نصیرالدین طوسی، تهران، ایران
چکیده
در دنیای امروز، هوش مصنوعی به عنوان یک عامل قدرتمندِ غیرقابل انکار در تمام امور انسانی ورود پیدا کرده است. این هوش با استفاده از الگوریتم‌های شبکه عصبی و کلان داده‌ها آموزش دیده و پیش‌بینی می‌کند. در گذر زمان انواع متفاوتی از شبکه‌های عصبی برای کاربردهای متفاوت معرفی و توسعه داده شده که یکی از اینها، شبکه عصبی بازگشتی (RNN) است. مدل RNN بخاطر معماری و ساختار آن، عملکرد قابل قبولی بر روی داده های سری زمانی (Time series) – داده‌هایی که در آن ویژگی های مختلف از یک پدیده در گام های زمانی ثابت برداشت شده- دارد. در حمل و نقل موارد زیادی از داده های سری زمانی وجود دارد مانند: حجم عبوری ترافیک از یک نقطه خاص در بازه های زمانی ثایت، تعداد و جنسیت و ... مسافرین مترو در ساعات مختلف شبانه‌روز، ویژگی‌های حرکت مثل موقعیت و سرعت و شتابِ یک عامل ترافیکی مثل عابر پیاده رو هر لحظه در پیاده رو. مورد آخر که سوابق و تاریخچۀ ویژگی‌های حرکتیِ عابر است، اساس کار این پژوهش است. در این مقاله با استفاده از 3 زیرمجموعۀ شبکه عصبی بازگشتی یعنی مدل‌های Vanilla LSTM، Stacked LSTM و GRU به پیش‌بینی مسیر حرکت عابر می‌پردازیم. 2 هدف اصلی این پژوهش اولاً تخمین موقعیت آینده یک عابر جهت شناسایی و رفع شرایط خطرآفرین در تعامل عابر و سیستم خودران (مثل ربات کالارسان) در پیاده رو و ثانیاً بررسی عملکرد این 3 مدل در داده های سری زمانی است. نتایج نشان داد که در افق زمانی پیش‌بینی کوتاه مدت، مدل GRU عملکرد بهتری نسبت به دیگر مدل‌ها دارد. اما با افزایش افق زمانی پیش‌بینی و افزایش پیچیدگی‎های داده‎‌ها، مدل Stacked LSTM عملکرد بهتری نسبت به سایرین دارد.
کلیدواژه‌ها

عنوان مقاله English

Investigating the performance of various types of Recurrent Neural Networks in predicting time series data in transportation; data type: pedestrian movement path on the sidewalk

نویسندگان English

Ali Edrisi 1
Mohammad Zahedi 2
1 Associate Professor, Department of Civil Engineering, K.N. Toosi University of Technology, Tehran, Iran
2 M.Sc., Student, Department of Civil Engineering, K.N. Toosi University of Technology, Tehran, Iran
چکیده English

In today's world, artificial intelligence has entered as an undeniable powerful factor in all human affairs. This intelligence is trained and predicted using neural network algorithms and big data. Over time, different types of neural networks have been introduced and developed for different applications, one of which is the Recurrent Neural Network (RNN). Due to its architecture and structure, the RNN model has an acceptable performance on time series data - data in which different features of a phenomenon are taken at fixed time steps. In transportation, there are many cases of time series data such as: the volume of traffic passing through a specific point in fixed time intervals, the number and gender, etc. of subway passengers at different times of the day and night, movement characteristics such as the position and speed and acceleration of a traffic agent such as a pedestrian at any moment on the sidewalk. The last case, which is the records and history of pedestrian movement characteristics, is the basis of this research. In this paper, we use 3 subsets of RNN, namely Vanilla LSTM, Stacked LSTM, and GRU models, to predict the Trajectory of a pedestrian. main goals of this research is to first estimate the future position of a pedestrian in order to identify and eliminate hazardous conditions in the interaction between a pedestrian and an autonomous system (such as a delivery robot) on the sidewalk, and secondly to examine the performance of these 3 models on time series data. The results showed that in the short-term prediction time horizon, the GRU model performs better than the other models. However, with increasing the prediction time horizon and increasing the complexity of the data, the Stacked LSTM model performs better than the others.

کلیدواژه‌ها English

Trajectory Prediction, Time Series Data, Recurrent Neural Network, LSTM, GRU
-Amalou, I., Mouhni, N., & Abdali, A. (2022). Multivariate time series prediction by RNN architectures for energy consumption forecasting. Energy Reports, 8, 1084-1091.
- Bock, J., Krajewski, R., Moers, T., Runde, S., Vater, L., & Eckstein, L. (2020, October). The ind dataset: A drone dataset of naturalistic road user trajectories at german intersections. In 2020 IEEE Intelligent Vehicles Symposium (IV), IEEE. 1929-1934.
-Cahuantzi, R., Chen, X., & Güttel, S. (2023, July). A comparison of LSTM and GRU networks for learning symbolic sequences. In Science and Information Conference, 771-785. Cham: Springer Nature Switzerland.
- Feng, W., Guan, N., Li, Y., Zhang, X., & Luo, Z. (2017). Audio visual speech recognition with multimodal recurrent neural networks. In 2017 International Joint Conference on Neural Networks (IJCNN), IEEE. 681-688.
- Gruel, W., & Stanford, J. M. (2016). Assessing the long-term effects of autonomous vehicles: a speculative approach. Transportation Research Procedia, 13, 18-29.
- Hansson, S. O., Belin, M. Å., & Lundgren, B. (2021). Self-driving vehicles—an ethical overview. Philosophy & Technology, 34(4), 1383-1408.
- Hu, J., Wang, X., Zhang, Y., Zhang, D., Zhang, M., & Xue, J. (2020). Time series prediction method based on variant LSTM recurrent neural network. Neural Processing Letters, 52(2), 1485-1500.
- Ji, X. (2018, April). The impact of self-driving cars on existing transportation networks. In AIP Conference Proceedings, Vol. 1955, No. 1, 040142. AIP Publishing LLC.
- Liang, Y., Wen, H., Nie, Y., Jiang, Y., Jin, M., Song, D. & Wen, Q. (2024). Foundation models for time series analysis: A tutorial and survey. In Proceedings of the 30th ACM SIGKDD conference on knowledge discovery and data mining, 6555-6565.
- Liu, P., Yang, R., & Xu, Z. (2019). How safe is safe enough for self‐driving vehicles? Risk analysis, 39(2), 315-325.
- Saleh, K., Hossny, M., & Nahavandi, S. (2017, October). Intent prediction of vulnerable road users from motion trajectories using stacked LSTM network. In 2017 IEEE 20th International Conference on intelligent transportation systems (ITSC), IEEE. 327-332.
- Salehinejad, H., Sankar, S., Barfett, J., Colak, E., & Valaee, S. (2017). Recent advances in recurrent neural networks. arXiv preprint arXiv:1801.01078.
- Shih, S. Y., Sun, F. K., & Lee, H. Y. (2019). Temporal pattern attention for multivariate time series forecasting. Machine Learning, 108(8), 1421-1441.
- Torres, J. F., Hadjout, D., Sebaa, A., Martínez-Álvarez, F., & Troncoso, A. (2021). Deep learning for time series forecasting: a survey. Big data, 9(1), 3-21.
- Weerakody, P. B., Wong, K. W., Wang, G., & Ela, W. (2021). A review of irregular time series data handling with gated recurrent neural networks. Neurocomputing, 441, 161-178.
- Yamak, P. T., Yujian, L., & Gadosey, P. K. (2019, December). A comparison between arima, lstm, and gru for time series forecasting. In Proceedings of the 2019 2nd International Conference on Algorithms, Computing and Artificial Intelligence, 49-55.
- Zhao, W., Alwidian, S., & Mahmoud, Q. H. (2022). Adversarial training methods for deep learning: A systematic review. Algorithms, 15(8), 283.
-­Alahi, A., Goel, K., Ramanathan, V., Robicquet, A., Fei-Fei, L., & Savarese, S. (2016). Social lstm: Human trajectory prediction in crowded spaces. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 961-971.
-­Bartoli, F., Lisanti, G., Ballan, L., & Del Bimbo, A. (2018). Context-aware trajectory prediction. In 2018 24th International Conference on Pattern Recognition (ICPR) IEEE, 1941-1946.
-­Bighashdel, A., & Dubbelman, G. (2019). A survey on path prediction techniques for vulnerable road users: From traditional to deep-learning approaches. In 2019 IEEE intelligent Transportation Systems Conference (ITSC) IEEE.1039-1046.
-­Huynh, M., & Alaghband, G. (2019). Trajectory prediction by coupling scene-LSTM with human movement LSTM. In International Symposium on Visual Computing, 244-259. Cham: Springer International Publishing.
-­Jiang, X., Lin, W., & Liu, J. (2019). A method of pedestrian trajectory prediction based on LSTM. In Proceedings of the 2019 2nd International Conference on Computational Intelligence and Intelligent Systems, 79-84.
-­Kumar, B. P., & Hariharan, K. (2020). Multivariate time series traffic forecast with long short-term memory based deep learning model. In 2020 International conference on power, instrumentation, control and computing (PICC)­, IEEE. 1-5.
-Nosouhian, S., Nosouhian, F., & Khoshouei, A. K. (2021). A review of recurrent neural network architecture for sequence learning: Comparison between LSTM and GRU.
-­Rudenko, A., Palmieri, L., Herman, M., Kitani, K. M., Gavrila, D. M., & Arras, K. O. (2020). Human motion trajectory prediction: A survey. The International Journal of Robotics Research, 39(8), 895-935.
-­Simon, H. (2009). Neural networks and learning machines. Pearson International Edition, 282-283.
-­Sun, Q., Jankovic, M. V., Bally, L., & Mougiakakou, S. G. (2018, November). Predicting blood glucose with a lstm and bi-lstm based deep neural network. In 2018 14th symposium on neural networks and applications (NEUREL), IEEE. 1-5.