Joint Out-of-Distribution Detection and Uncertainty Estimation for Trajectory Prediction

TL;DR

This paper presents a combined approach for out-of-distribution detection and uncertainty estimation in trajectory prediction for automated driving, improving reliability assessment in both novel and complex in-distribution scenarios.

Contribution

It introduces two modules integrated with an encoder-decoder network to detect OOD samples and estimate prediction errors, advancing trajectory prediction reliability.

Findings

Outperforms prior work in OOD detection by 2.8%.

Achieves 10.1% improvement in uncertainty estimation.

Effective on the Shifts dataset.

Abstract

Despite the significant research efforts on trajectory prediction for automated driving, limited work exists on assessing the prediction reliability. To address this limitation we propose an approach that covers two sources of error, namely novel situations with out-of-distribution (OOD) detection and the complexity in in-distribution (ID) situations with uncertainty estimation. We introduce two modules next to an encoder-decoder network for trajectory prediction. Firstly, a Gaussian mixture model learns the probability density function of the ID encoder features during training, and then it is used to detect the OOD samples in regions of the feature space with low likelihood. Secondly, an error regression network is applied to the encoder, which learns to estimate the trajectory prediction error in supervised training. During inference, the estimated prediction error is used as the uncertainty. In our experiments, the combination of both modules outperforms the prior work in OOD detection and uncertainty estimation, on the Shifts robust trajectory prediction dataset by $2.8 \%$ and $10.1 \%$, respectively. The code is publicly available.