DeepAI AI Chat
Log In Sign Up

An Empirical Bayes Analysis of Vehicle Trajectory Models

by   Yue Yao, et al.

We present an in-depth empirical analysis of the trade-off between model complexity and representation error in modelling vehicle trajectories. Analyzing several large public datasets, we show that simple linear models do represent realworld trajectories with high fidelity over relevant time scales at very moderate model complexity. This finding allows the formulation of trajectory tracking and prediction as a Bayesian filtering problem. Using an Empirical Bayes approach, we estimate prior distributions over model parameters from the data that inform the motion models necessary in the trajectory tracking problem and that can help regularize prediction models. We argue for the use of linear models in trajectory prediction tasks as their representation error is much smaller than the typical epistemic uncertainty in this task.


MEAT: Maneuver Extraction from Agent Trajectories

Advances in learning-based trajectory prediction are enabled by large-sc...

Representing Spatial Trajectories as Distributions

We introduce a representation learning framework for spatial trajectorie...

DeepTrack: Lightweight Deep Learning for Vehicle Path Prediction in Highways

Vehicle trajectory prediction is an essential task for enabling many int...

Quantifying the Complexity of Standard Benchmarking Datasets for Long-Term Human Trajectory Prediction

Methods to quantify the complexity of trajectory datasets are still a mi...

Visualization of Model Parameter Sensitivity along Trajectories in Numerical Weather Predictions

Numerical weather prediction models rely on parameterizations for subgri...

Inertial Single Vehicle Trajectory Prediction Baselines and Applications with the NGSIM Dataset

In the recent vehicle trajectory prediction literature, the most common ...

Adaptive α Significance Level for Linear Models

We put forward an adaptive alpha that decreases as the information grows...