Trajectron++: Dynamically-Feasible Trajectory Forecasting with Heterogeneous Data

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-ofthe-art deterministic and generative methods.

A Review of Deep Learning-Based Methods for Pedestrian Trajectory Prediction

“…In [82], T. Salzmann et al proposed a method that forecast the futures-conditional trajectories of the general number of agents (i.e., pedestrians, vehicles) with distinct semantic classes, while including heterogeneous data. To encode agent interactions, they developed a system in which each agent has a semantic class namely car, bus or pedestrian and provides information about their position histories with context size, spatial resolution, and semantic channels.…”

Vehicle Trajectory Prediction Using Generative Adversarial Network With Temporal Logic Syntax Tree Features

“…However, since no class information is provided during the process of encoding, the model can not learn class conditional trajectory patterns but instead the averaged trajectories of different classes. Several graph attention networks [9,12] have incorporated class information into the model by concatenating class information with other inputs and applying embedding to extract features. Nevertheless, no experiment results directly demonstrate how the addition of class information impact the pre-diction performance.…”

Limits of Probabilistic Safety Guarantees when Considering Human Uncertainty

“…Safety Threshold Gaussian Process [2], [10]- [12] δ ≥ 0.001 Gaussian Uncertainty w/ Dynamics [13]- [15] δ ≥ 0.001 Bayesian NN [5], [16] δ ≥ 0.05 Noisy Rational Model [3] δ ≥ 0.01 Hidden Markov Model [17], [18] δ ≥ 0.01 Quantile Regression [5] δ ≥ 0.05 Scenario Optimization [19]- [21] δ ≥ 0.01 Generative Models (e.g. GANs) [9], [22], [23] N/A • Gaussian Process (GP): These approaches model other agents' trajectories as Gaussian processes, which treat trajectory uncertainty as a multivariate Gaussian [2], [11], [12], [24]. There are several extensions, such as the IGP model [25] (which accounts for interaction between multiple agents), or others [26], [27].…”