DeepAI AI Chat
Log In Sign Up

Differentiable Hybrid Traffic Simulation

by   Sanghyun Son, et al.

We introduce a novel differentiable hybrid traffic simulator, which simulates traffic using a hybrid model of both macroscopic and microscopic models and can be directly integrated into a neural network for traffic control and flow optimization. This is the first differentiable traffic simulator for macroscopic and hybrid models that can compute gradients for traffic states across time steps and inhomogeneous lanes. To compute the gradient flow between two types of traffic models in a hybrid framework, we present a novel intermediate conversion component that bridges the lanes in a differentiable manner as well. We also show that we can use analytical gradients to accelerate the overall process and enhance scalability. Thanks to these gradients, our simulator can provide more efficient and scalable solutions for complex learning and control problems posed in traffic engineering than other existing algorithms. Refer to for our project.


page 1

page 6

page 7

page 8


NeuralSim: Augmenting Differentiable Simulators with Neural Networks

Differentiable simulators provide an avenue for closing the sim-to-real ...

Dojo: A Differentiable Simulator for Robotics

We present a differentiable rigid-body-dynamics simulator for robotics t...

DiffCloth: Differentiable Cloth Simulation with Dry Frictional Contact

Cloth simulation has wide applications in computer animation, garment de...

Fast Traffic Engineering by Gradient Descent with Learned Differentiable Routing

Emerging applications such as the metaverse, telesurgery or cloud comput...

Open Traffic Models – A framework for hybrid simulation of transportation networks

This paper introduces a new approach to hybrid traffic modeling, along w...

CAC and Traffic Modeling for Integrated Macrocell/Femtocell Networks

Dense femtocells and the integration of these femtocells with the macroc...