Off-road Autonomy and Robotics
This research focuses on developing a systematic data-driven analytical and computational framework for the navigation of autonomous vehicles in the off-road environment. The data-driven tools are based on linear operator theory involving Koopman and Perron-Frobenius operators. The developed framework is tested on an experimental platform consisting of an F1TENTH vehicle, 1/5th scale Hunter SE. We are currently implementing the framework for full-scale Warthog and MRZR vehicles.
The linear operator theoretical framework is also being applied to the autonomous control of quadruped robots. The operator’s theoretical tools are used to solve the planning problems with safety constraints and real-time control problems. One of our main contributions includes the analytical construction of density functions for safe navigation in static and dynamic environments, which consist of dynamically changing unsafe sets and targets.
The density function is based on the occupancy-based physical interpretation of safety constructed using linear operator theory. The control problem is solved using Koopman-based Model Predictive Control, where we proposed using a parameterized family of switched Koopman models that can adapt to changing terrain properties and environmental conditions.
Selected publications
- A. Joglekar, C. Samak, T. Samak, V. Krovi, U. Vaidya, Expanding Autonomous Ground Vehicle Navigation Capabilities through a Multi-Model Parameterized Koopman Framework.
- C. Samak, T. Samak, A. Joglekar, U. Vaidya, V. Krovi, Digital Twins Meet the Koopman Operator: Data-Driven Learning for Robust Autonomy.
- J. Moyalan, SSKS Narayanan, A. Zheng, U. Vaidya, Synthesizing Controller for Safe Navigation using Control Density Function, IEEE American Control Conference (ACC) 2024.
- A. Joglekar, S. Sutavani, C. Samak, T. Samak, KC Kosaraju, J. Smereka, D. Gorsich, U. Vaidya, V. Krovi, Data-Driven Modeling and Experimental Validation of Autonomous Vehicles Using Koopman Operator, IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) 2023.
- SSKS Narayanan, D. Tellez-Castro, S. Sutavani, U. Vaidya, SE (3) Koopman-MPC: Data-driven Learning and Control of Quadrotor UAVs, Modeling, Estimation and Control Conference MECC 2023.
- A. Zheng, SSKS Narayanan, U. Vaidya, Safe navigation using density functions, IEEE Robotics and Automation Letters 2023.
- J. Moyalan, Y. Chen, U. Vaidya, Convex Approach to Data-driven Off-road Navigation via Linear Transfer Operators, IEEE Robotics and Automation Letters 2023.
