Platform Integration and GUI with Project Varuna

Staying true to our mission, Project Varuna simplifies access to our cutting-edge research pipeline no extensive dependencies, coding knowledge, or complex setup required. Our framework enables users to leverage our end-to-end autonomy pipelines across multiple simulation and hardware platforms with just a few clicks.

Multi-Scale & Multi-Platform Autonomy

Project Varuna is designed from the ground up to provide a flexible, scalable, data-driven framework for end-to-end modeling, motion planning, and control.

Our software packages empower users to generate custom models for platforms of varying sizes and capabilities using a standardized data format. With built-in GUI, users can map ROS topics directly, allowing seamless integration with diverse sensors and control systems.

This framework has been rigorously tested across multiple platforms, including F1Tenth (1/10th scale RC car) and Hunter SE (1/5th scale Ackermann-steered vehicle equipped with LiDAR, IMU, and GNSS) in simulation environments like AutoDRIVE and ISAAC Sim, as well as real-world hardware deployments.

User-Friendly GUI and Executable Packages

To ensure ease of use and accessibility, the entire pipeline is compiled into a single executable eliminating the complexity of managing dependencies, codebases, and lengthy parameter configurations.

Paired with the intuitive front-end GUI, this user-friendly interface allows for rapid generation of MMPK models and real-world deployment in under 2 minutes. It’s incredibly simple!

Just follow these three steps through the GUI:

Select training data or pretrained models, test data, and platform.
Choose your motion planner, adjust MPC penalties, and configure ROS topics based on the platform.
Pick your desired path, and you're ready to go!