Energy-based Potential Games for Joint Motion Forecasting and Control
This work addresses interaction modeling for automated driving, offering an incremental improvement by integrating game theory with neural networks for better interpretability and performance.
The paper tackles the challenge of unknown game parameters in applying game theory to multi-agent motion forecasting and control for robotics, such as automated driving, by proposing an Energy-based Potential Game formulation and an end-to-end learning method that improves predictive performance on simulations and real-world datasets.
This work uses game theory as a mathematical framework to address interaction modeling in multi-agent motion forecasting and control. Despite its interpretability, applying game theory to real-world robotics, like automated driving, faces challenges such as unknown game parameters. To tackle these, we establish a connection between differential games, optimal control, and energy-based models, demonstrating how existing approaches can be unified under our proposed Energy-based Potential Game formulation. Building upon this, we introduce a new end-to-end learning application that combines neural networks for game-parameter inference with a differentiable game-theoretic optimization layer, acting as an inductive bias. The analysis provides empirical evidence that the game-theoretic layer adds interpretability and improves the predictive performance of various neural network backbones using two simulations and two real-world driving datasets.