Bayesian Prediction of Future Street Scenes using Synthetic Likelihoods
This addresses the need for accurate and calibrated scene anticipation in autonomous driving, representing a novel method for a known bottleneck rather than an incremental improvement.
The paper tackled the problem of predicting uncertain and multi-modal future street scenes for autonomous agents, proposing a Bayesian formulation with synthetic likelihoods that encourages diverse models and achieves state-of-the-art predictions with calibrated probabilities on the Cityscapes dataset.
For autonomous agents to successfully operate in the real world, the ability to anticipate future scene states is a key competence. In real-world scenarios, future states become increasingly uncertain and multi-modal, particularly on long time horizons. Dropout based Bayesian inference provides a computationally tractable, theoretically well grounded approach to learn likely hypotheses/models to deal with uncertain futures and make predictions that correspond well to observations -- are well calibrated. However, it turns out that such approaches fall short to capture complex real-world scenes, even falling behind in accuracy when compared to the plain deterministic approaches. This is because the used log-likelihood estimate discourages diversity. In this work, we propose a novel Bayesian formulation for anticipating future scene states which leverages synthetic likelihoods that encourage the learning of diverse models to accurately capture the multi-modal nature of future scene states. We show that our approach achieves accurate state-of-the-art predictions and calibrated probabilities through extensive experiments for scene anticipation on Cityscapes dataset. Moreover, we show that our approach generalizes across diverse tasks such as digit generation and precipitation forecasting.