System Identification of Thrust and Torque Characteristics for a Bipedal Robot with Integrated Propulsion
Thomas Cahill
TL;DR
This work advances thruster-assisted bipedal robotics by developing and validating a comprehensive approach to thrust and torque identification on Harpy, a lightweight biped with integrated EDF propulsion. It presents a theory-grounded thrust equation, CFD validation via ANSYS, and tethered bench testing, coupled with three independent methods for determining the joint motor torque constant $K_t$, including a final sensor-based estimate $K_t=0.0311$ Nm/A. The study demonstrates hardware integration with carefully designed mounts, leverages impedance-based control to enable force-responsive behavior, and offers a clear roadmap for improving stability, robustness, and dynamic performance in hybrid ground-air locomotion. Collectively, these contributions enable rapid thrust design, accurate torque modeling, and improved control strategies for agile, terrain-adaptive robots in real-world environments.
Abstract
Bipedal robots represent a remarkable and sophisticated class of robotics, designed to emulate human form and movement. Their development marks a significant milestone in the field. However, even the most advanced bipedal robots face challenges related to terrain variation, obstacle negotiation, payload management, weight distribution, and recovering from stumbles. These challenges can be mitigated by incorporating thrusters, which enhance stability on uneven terrain, facilitate obstacle avoidance, and improve recovery after stumbling. Harpy is a bipedal robot equipped with six joints and two thrusters, serving as a hardware platform for implementing and testing advanced control algorithms. This thesis focuses on characterizing Harpy's hardware to improve the system's overall robustness, controllability, and predictability. It also examines simulation results for predicting thrust in propeller-based mechanisms, the integration of thrusters into the Harpy platform and associated testing, as well as an exploration of motor torque characterization methods and their application to hardware in relation to closed-loop force-based impedance control.
