Trajectron++: Dynamically-Feasible Trajectory Forecasting With Heterogeneous Data¶
Authors: Tim Salzmann, Boris Ivanovic, Punarjay Chakravarty, Marco Pavone
Published: 2020 (Conference Paper)
Source: European Conference on Computer Vision (ECCV)
Algorithm: Trajectron++
arXiv: 2001.03093
DOI: 10.1007/978-3-030-58523-5_40
Summary¶
Extends Trajectron with dynamic feasibility constraints and heterogeneous input data (HD maps, agent types), using a CVAE-based graph recurrent network to produce multi-modal trajectory distributions for multiple interacting agents simultaneously.
Abstract¶
Reasoning about human motion is an important prerequisite to safe and socially-aware robotic navigation. As a result, multi-agent behavior prediction has become a core component of modern human-robot interactive systems, such as self-driving cars. While there exist many methods for trajectory forecasting, most do not enforce dynamic constraints and do not account for environmental information (e.g., maps). Towards this end, we present Trajectron++, a modular, graph-structured recurrent model that forecasts the trajectories of a general number of diverse agents while incorporating agent dynamics and heterogeneous data (e.g., semantic maps). Trajectron++ is designed to be tightly integrated with robotic planning and control frameworks; for example, it can produce predictions that are optionally conditioned on ego-agent motion plans. We demonstrate its performance on several challenging real-world trajectory forecasting datasets, outperforming a wide array of state-of-the-art deterministic and generative methods.
Links¶
Primary
Standard
Alternate
Tags¶
-
Trajectory prediction
-
Motion forecasting
-
Multi-agent prediction
-
Graph neural networks
-
Heterogeneous data
-
Probabilistic prediction
-
Conditional variational autoencoder