Home > Robotics > Euro-PVI: Pedestrian Vehicle Interactions in Dense Urban Centers

Euro-PVI: Pedestrian Vehicle Interactions in Dense Urban Centers

Dense inner city environments are one of the most challenging locations for self-driving vehicles. Most of the current trajectory prediction datasets are collected on multi-lane roads, where interactions between vehicles and pedestrians or bicyclists are scarce.

A recent study on arXiv.org introduces a dataset containing a rich and diverse set of interactions between ego-vehicle and pedestrians.

Image credit: Sionk via Wikimedia (CC BY-SA 4.0)

It was collected near busy urban landmarks in two Belgium cities. A novel Joint-β-Conditional Variational Autoencoder models a “shared” latent space between agents to better capture the effect of interactions in the latent space and accurately represent the multi-modal distribution of trajectories. State-of-the-art results were demonstrated with this approach for the interaction prediction task even when using a more challenging novel dataset.

Accurate prediction of pedestrian and bicyclist paths is integral to the development of reliable autonomous vehicles in dense urban environments. The interactions between vehicle and pedestrian or bicyclist have a significant impact on the trajectories of traffic participants e.g. stopping or turning to avoid collisions. Although recent datasets and trajectory prediction approaches have fostered the development of autonomous vehicles yet the amount of vehicle-pedestrian (bicyclist) interactions modeled are sparse. In this work, we propose Euro-PVI, a dataset of pedestrian and bicyclist trajectories. In particular, our dataset caters more diverse and complex interactions in dense urban scenarios compared to the existing datasets. To address the challenges in predicting future trajectories with dense interactions, we develop a joint inference model that learns an expressive multi-modal shared latent space across agents in the urban scene. This enables our Joint-β-cVAE approach to better model the distribution of future trajectories. We achieve state of the art results on the nuScenes and Euro-PVI datasets demonstrating the importance of capturing interactions between ego-vehicle and pedestrians (bicyclists) for accurate predictions.

Research paper: Bhattacharyya, A., Olmeda Reino, D., Fritz, M., and Schiele, B., “Euro-PVI: Pedestrian Vehicle Interactions in Dense Urban Centers”, 2021. Link: https://arxiv.org/abs/2106.12442


Notify of
Inline Feedbacks
View all comments
Would love your thoughts, please comment.x