Collaborative Humanoid Operational Work Robot (CHOWbot)

Our Approach We propose using a robotic humanoid upper torso system of our design, and common kitchen tools in combination with novel control behaviors to perform back-of-the-house duties in an unstructured environment. Our approach represents a significant improvement from state-of-the-art systems, developing a two-armed manipulator system adaptable to the chaotic spaces of the world at a cost equal to or less than a traditional industrial robot. Our Deliverable We propose simulating a two-armed, upper torso humanoid of our Work-Class Perseus effort. We have hardware customers who use our Perseus system for Artificial Intelligence development that would team with our controls effort. Our team would dive directly into controllability studies for the unique and varied tasks and environment of an industrial kitchen. We anticipate that initial tele-operated studies provide proof-of-concept returns which can be de-risked during Phase I before being fully developed into hardware and autonomous behaviors during a Phase II effort. Our concept incorporates and combines previous hardware and controls work of ours in a novel way, allowing a significant impact in modelling and delivering a real-world robotic solution for the chaotic environment of a DFAC. The Rationale Behind Our Approach We have built early prototypes of our Work-Class Perseus Humanoid hardware, but every revision of the system improves the robustness and capability of our already reliable platform. Using this existing platform allows us to focus risk on the execution of tasks instead of the reliability of hardware. We have designed certain foundational control software to match perfectly with our propriety hardware, further reducing the risk pool. This risk reduction allows our controls team to get to work immediately to understand the back-of-house tasks and begin to break those challenges down to a number of simpler behaviors and model, beginning the process of modelling the various tasks from the ground up. Our team would develop a number of strategies in a remote or tele-operation style – proving the path forward to full autonomy during Phase I, allowing us to identify the most powerful, labor-saving, and efficiency improving solutions to take into full autonomy during Phase II. The technical path forward For our path forward, Apptronik in Phase II, will move the control algorithm designs to test and evaluation, shifting from human in the loop tele-operation, to supervision-optional autonomy. We would develop a prioritized function set through end-user dialog and starting with the most significant add autonomy to each behavior in turn down the list.