DINAMIK HUDUDLARDA OB’YEKT TRAYEKTORIYASINI ANIQLASH VA PROGNOZLASHNING MATEMATIK ASOSLARI
Keywords:
trayektoria prognozi, Bayesian model, Gaussian jarayon, LSTM, dinamik muhit, ADE/FDE mezonlari, aqlli transport, ob'yektlarni kuzatish, stoxastik dinamika, deterministik model.Abstract
Ushbu maqolada dinamik hududlarda harakatlanuvchi ob'yektlar trayektoriyasini aniqlash va kelajakdagi holatini prognozlashning matematik asoslari o'rganiladi. Tadqiqotning dolzarbligi shundaki, aqlli transport tizimlari, robototexnika va xavfsizlik monitoringida ob'yektning kelgusi holatini oldindan bilish — to'qnashuvlarni oldini olish, trafik boshqaruvi va avtomatik rejalashtirish uchun muhimdir. Ammo dinamik muhitlarda — o'zgaruvchan tezlik, noto'g'ri o'lchovlar va ko'p ob'yektning o'zaro ta'siri — prognoz aniqligini saqlash qiyin masala bo'lib qolmoqda. Muammoni hal etish uchun Bayesian trayektoria modeli, Gaussian jarayonlar (GP) regressiyasi va Long Short-Term Memory (LSTM) neyron tarmog'ini birlashtiradigan gibrid prognoz tizimi taklif etiladi. ETH/UCY pedestrian va nuScenes transport ma'lumotlar to'plamlarida o'tkazilgan tajribalar 1 soniya ufqda ADE = 0.28 m va FDE = 0.41 m xato darajasiga erishganligini ko'rsatdi. Ilmiy yangilik — deterministik Bayesian model, ehtimollik GP regressiyasi va chuqur o'rganish LSTM ni birlashtirgan uch qatlamli gibrid prognoz arxitekturasini yaratishdan iborat.
References
[1] Rudenko A. et al. Human motion trajectory prediction: A survey // The International Journal of Robotics Research. — 2020. — Vol. 39, No. 8. — P. 895–935. DOI: 10.1177/0278364920917446
[2] Lefevre S., Vasquez D., Laugier C. A survey on motion prediction and risk assessment for intelligent vehicles // ROBOMECH Journal. — 2014. — Vol. 1, No. 1. — P. 1–14. DOI: 10.1186/s40648-014-0001-z
[3] Schöller C. et al. What the constant velocity model can teach us about pedestrian motion prediction // IEEE Robot. Autom. Lett. — 2020. — Vol. 5, No. 2. — P. 1696–1703. DOI: 10.1109/LRA.2020.2969925
[4] Liang J. et al. The garden of forking paths: Towards multi-future trajectory prediction // Proc. IEEE CVPR. — 2020. — P. 10508–10518. DOI: 10.1109/CVPR42600.2020.01052
[5] Huang Z. et al. LSTM based trajectory prediction model for cyclist considering interaction with other participants // Pattern Recognition. — 2022. — Vol. 123. — P. 108392. DOI: 10.1016/j.patcog.2021.108392
[6] Kalman R.E. A new approach to linear filtering and prediction problems // Trans. ASME J. Basic Eng. — 1960. — Vol. 82, No. 1. — P. 35–45. DOI: 10.1115/1.3662552
[7] Singer R.A. Estimating optimal tracking filter performance for manned maneuvering targets // IEEE Trans. Aerosp. Electron. Syst. — 1970. — Vol. 6, No. 4. — P. 473–483. DOI: 10.1109/TAES.1970.310128
[8] Titsias M., Lawrence N.D. Bayesian Gaussian process latent variable model // Proc. AISTATS. — 2010. — Vol. 9. — P. 844–851.
[9] Rasmussen C.E., Williams C.K.I. Gaussian Processes for Machine Learning. — MIT Press, 2006. — 248 p.
[10] Helbing D., Molnar P. Social force model for pedestrian dynamics // Physical Review E. — 1995. — Vol. 51, No. 5. — P. 4282–4286. DOI: 10.1103/PhysRevE.51.4282
[11] Alahi A. et al. Social LSTM: Human trajectory prediction in crowded spaces // Proc. IEEE CVPR. — 2016. — P. 961–971. DOI: 10.1109/CVPR.2016.110
