Closed-loop Simulation

InterSim is born for closed-loop interactive simulations.
Closed-loop simulations are pivotal to revealing failure patterns of predictors and planners before their deployment.
InterSim leverages relation reasoning models learned from a real-driving dataset to simulate human drivers' behaviors.
It is the best simulator to test your predictor and planner for L4 urban autonomous driving.


Forget about complex versions and directories configurations.
InterSim dashboard helps you to focus on what is important with diverse default and customized settings.
You can upload your predictors and planners, select the configs, launch a simulation and check them with our dashboard. Here are some simulation results with playbacks from the Waymo Open Motion Dataset.


Interactive traffic simulation is crucial to autonomous driving systems by enabling testing for planners in a more scalable and safe way compared to real-world road testing. Existing approaches learn an agent model from large-scale driving data to simulate realistic traffic scenarios, yet it remains an open question to produce consistent and diverse multi-agent interactive behaviors in crowded scenes. In this work, we present InterSim, an interactive traffic simulator for testing autonomous driving planners. Given a test plan trajectory from the ego agent, InterSim reasons about the interaction relations between the agents in the scene and generates realistic trajectories for each environment agent that are consistent with the relations. We train and validate our model on a large-scale interactive driving dataset. Experiment results show that InterSim achieves better simulation realism and reactivity in two simulation tasks compared to a state-of-the-art learning-based traffic simulator.

Interactive Closed-loop Simulation Pipeline

Illustration of InterSim. In this example, given a new plan for the ego agent (in cyan) to slow down, the simulator updates its simulated trajectories for the environment agents as follows. First, it checks for potential collisions with all environment agents and labels colliding ones as the relevant agents in yellow. For each relevant agent, such as Env #1, it predicts the interaction relation and updates its trajectory based on the relation using a goal driven trajectory predictor. Second, it resolves collisions between the newly updated trajectories of the environment agent(s) and the remaining agents (i.e. Env #2) iteratively until all collisions are resolved. In the end, InterSim successfully generates scene consistent trajectories for Env #1 and Env #2 to react and slow down, and commit these trajectories to simulate for the next step.

Efficient and Realistic Closed-loop Simulations

InterSim generates proper reactions to a plan slightly different from the playback of the Waymo Open Motion Dataset. Each scenario lasts for eight seconds.

Play with More Demos

This visualization loops through 30 randomly chosen simulations with the Waymo Open Motion Dataset - Interactive Validation Subset. The white vehicle is controlled by a fairly basic planner. The yellow agents, if any, were detected as relevant agents and hence controlled by the simulator. The red pin is the goal point assigned to the ego vehicle.

Drag to move the center of the visualization in case of misalignment.

Contact Us

We release InterSim Beta as an open-source project. We hope this tool can help the community push the frontline of interactive planning systems. If you are enlightened, please consider citing our paper. If you like our idea and want to give your planner a test, follow our easier-than-you-thought tutorials and fork our codes. Moreover, building a realistic planning simulator is a big challenge demanding broad collaborations across multiple communities. If you are annoyed by some bugs, unfinished functions, or lacking fundamental extensions, please consider contributing your effort and building it for all. And you are more than welcome to connect with us if you don't know where to start.


title={{InterSim}: Interactive Traffic Simulation via Explicit Relation Modeling},
author={Sun, Qiao and Huang, Xin and Williams, Brian and Zhao, Hang},
booktitle={2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)},