Award

HEFESTUS (Humanoid Enhancement Framework for Efficient and Safe Task-oriented Unmanned Sustainment) is an advanced robotics project aimed at transforming sustainment tasks within the United States Air Force. By developing the humanoid robot Hephaestus, we seek to enhance efficiency, safety, and readiness in sustainment operations. This project will leverage cutting-edge technologies in control algorithms, learning from demonstration (LfD), and behavior tree-based mission autonomy to create a robust, versatile robotic platform capable of performing complex tasks autonomously.Hephaestus will be designed to maximize its manipulation workspace through kinematic optimization techniques and quick tool-changing mechanisms, enabling it to handle a wide array of tasks. The learning framework will utilize teleoperated demonstrations to train generative models for motion primitives and skill coordination, ensuring the robot can perform intricate tasks with high precision and adaptability. The integration of advanced whole-body Model Predictive Control (MPC) algorithms will enable precise and dynamic task execution, enhancing the robot's athletic intelligence.The project will be organized into four main thrusts: (1) Hardware for Maximum Workspace, (2) Learning from Demonstration for Task Intelligence, (3) Whole-Body Athletic Intelligence, and (4) Behavior Trees for Mission Intelligence. These efforts will culminate in a fully functional Hephaestus prototype capable of performing tasks such as using tools like sanders and drills, pushing carts loaded with parts, and loading and unloading parts for maintenance operations.The final demonstration will be conducted at Warner Robins Air Logistics Complex (WR-ALC), showcasing Hephaestus's ability to navigate the facility and perform various sustainment tasks. This project aims to achieve a Technology Readiness Level (TRL) of 7 by the end of Phase II, setting the stage for further advancements and commercialization in Phase III.By integrating mission, task, and athletic intelligence, Hephaestus will significantly reduce the burden on human personnel, enhance safety, and improve the efficiency of depot-level maintenance operations. The innovative learning techniques and robust control systems developed in this project will also have wide-ranging applications in commercial sectors, driving innovation and economic growth.