Likelihood-Based Diverse Sampling for Trajectory Forecasting
This work provides an incremental improvement in trajectory forecasting for autonomous driving and robotics by enhancing the diversity and quality of predictions from existing probabilistic models.
This paper addresses the problem of generating diverse and high-quality trajectory samples from pre-trained probabilistic models like normalizing flows. The proposed Likelihood-Based Diverse Sampling (LDS) method produces a set of trajectories in one shot, optimizing for high likelihood and spatial separation. LDS outperforms state-of-the-art post-hoc neural diverse forecasting methods on two challenging benchmarks.
Forecasting complex vehicle and pedestrian multi-modal distributions requires powerful probabilistic approaches. Normalizing flows (NF) have recently emerged as an attractive tool to model such distributions. However, a key drawback is that independent samples drawn from a flow model often do not adequately capture all the modes in the underlying distribution. We propose Likelihood-Based Diverse Sampling (LDS), a method for improving the quality and the diversity of trajectory samples from a pre-trained flow model. Rather than producing individual samples, LDS produces a set of trajectories in one shot. Given a pre-trained forecasting flow model, we train LDS using gradients from the model, to optimize an objective function that rewards high likelihood for individual trajectories in the predicted set, together with high spatial separation among trajectories. LDS outperforms state-of-art post-hoc neural diverse forecasting methods for various pre-trained flow models as well as conditional variational autoencoder (CVAE) models. Crucially, it can also be used for transductive trajectory forecasting, where the diverse forecasts are trained on-the-fly on unlabeled test examples. LDS is easy to implement, and we show that it offers a simple plug-in improvement over baselines on two challenging benchmarks. Code is at: https://github.com/JasonMa2016/LDS