A State-Space Solution to the Estimation of Interacting Vehicle Trajectories with Deep Neural Networks and Variational Bayes Filtering - LAAS - Laboratoire d'Analyse et d'Architecture des Systèmes
Communication Dans Un Congrès Année : 2021

A State-Space Solution to the Estimation of Interacting Vehicle Trajectories with Deep Neural Networks and Variational Bayes Filtering

Résumé

This paper addresses the estimation of trajectories of interacting vehicles at a microscopic scale, as a prerequisite to their prediction for risk assessment. A state-space solution is investigated, where both the Markov hidden state (continuousvalued, which captures the joint histories of vehicles) and the measurements (low-dimensional and noisy) admit a vehiclewise structure. The vehicles' transition models are assumed independent of each other, time-and vehicle-invariant, and coequal to an "egocentric" prior dynamics pdf. To cope with the vehicles' interactions, this pdf is conditioned on the full state vector as the past time index, which imposes a centralized estimation/prediction of the fleet motion. The two fundamental pillars of the approach are developed: learning of a Gaussian mixture egocentric transition model by means of Deep Neural Networks; synthesis of a stochastic variational Bayes filtering algorithm which features a decentralized vehicle-wise structure but takes into account interactions. Tests on highway scenarios are presented.
Fichier principal
Vignette du fichier
klempka_danes.pdf (1.1 Mo) Télécharger le fichier
Origine Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04860497 , version 1 (31-12-2024)

Identifiants

Citer

Tristan Klempka, Patrick Danès. A State-Space Solution to the Estimation of Interacting Vehicle Trajectories with Deep Neural Networks and Variational Bayes Filtering. 2021 IEEE International Workshop of Electronics, Control, Measurement, Signals and their application to Mechatronics (ECMSM), Jun 2021, Liberec, Czech Republic. pp.1-7, ⟨10.1109/ECMSM51310.2021.9468863⟩. ⟨hal-04860497⟩
0 Consultations
0 Téléchargements

Altmetric

Partager

More