Unlocking Efficient Vehicle Dynamics Modeling via Analytic World Models

TL;DR

This paper introduces Analytic World Models (AWMs), leveraging differentiable simulation to enable end-to-end learning of predictive and prescriptive models for autonomous driving, enhancing decision-making beyond reactive control.

Contribution

The paper presents a novel approach to world modeling by integrating differentiable dynamics within end-to-end predictors, enabling efficient learning of odometry, planning, and inverse states.

Findings

AWMs effectively learn relative odometry and planning.

Differentiable simulation improves end-to-end learning efficiency.

AWMs enhance decision-making in autonomous driving scenarios.

Abstract

Differentiable simulators represent an environment's dynamics as a differentiable function. Within robotics and autonomous driving, this property is used in Analytic Policy Gradients (APG), which relies on backpropagating through the dynamics to train accurate policies for diverse tasks. Here we show that differentiable simulation also has an important role in world modeling, where it can impart predictive, prescriptive, and counterfactual capabilities to an agent. Specifically, we design three novel task setups in which the differentiable dynamics are combined within an end-to-end computation graph not with a policy, but a state predictor. This allows us to learn relative odometry, optimal planners, and optimal inverse states. We collectively call these predictors Analytic World Models (AWMs) and demonstrate how differentiable simulation enables their efficient, end-to-end learning. In autonomous driving scenarios, they have broad applicability and can augment an agent's decision-making beyond reactive control.