Responsible AI-Based Control of Unmanned Ground Vehicles in Severe Dynamic Terrain Environments

Project began Q1 2023.

In ARC projects #1.A82, 1.A85 & 4.A87, the analytical approaches on vehicle decoupling and interactive dynamics between the steering systems and electrically driven wheels of a vehicle with a fully electric driveline system were investigated, new vehicle mobility and energy efficiency performance indices were introduced, and accordingly agile controls were designed. As a part of those projects, agile hierarchical maneuver control system was also developed. In this project, we continue to advance the framework of vehicle mobility-maneuver-efficiency controls by developing a vehicle agile maneuver control system with the contextual metrics for turnability, stability, and handling performance (which were defined in ARC project #3.A88). Accordingly, a 3-year fundamental research with the following features and novelties is proposed:

1. Quantification of severe terrain and environmental conditions for control design purpose,
2. An intrinsic, single-layer, collaborative, and agile in-wheel motor (IWM) powertrains control,
3. An agile maneuverability control in response to severe stochastic, unstructured, and dynamically changing terrains and environments, and
4. A modular self-learning Artificial Intelligence (AI)-based control system that conforms the U.S. Department of Defense (DoD) AI ethical principles as being responsible, equitable, traceable, reliable, and governable (U.S. DoD 2022a, U.S. DoD 2022b).

In this project, the mathematical model and computer simulations of FED-Alpha equipped with in- wheel-motor (IWM) powertrains, which is named FED-Epsilon, will be used to research AI- based maneuverability control in severe environmental and terrain conditions.

#1.A113