Narrow-Path, Dynamic Walking Using Integrated Posture Manipulation and Thrust Vectoring

TL;DR

The paper tackles dynamic walking on narrow paths with a multimodal legged-aerial robot by integrating posture manipulation and thrust-vectoring. It introduces the Husky Reduced-Order Model (HROM) derived from an energy-based Euler-Lagrange formulation and a collocation-based, polynomial-approximation optimization controller to compute joint and thruster commands. The simulations show stable body attitude, forward progression, and thruster commands within limits, validating the approach for narrow-path locomotion. This work provides a framework for extending multimodal locomotion to challenging terrains like slacklines and narrow conduits, with future work focusing on higher-fidelity validation and real-world experimentation.

Abstract

This research concentrates on enhancing the navigational capabilities of Northeastern Universitys Husky, a multi-modal quadrupedal robot, that can integrate posture manipulation and thrust vectoring, to traverse through narrow pathways such as walking over pipes and slacklining. The Husky is outfitted with thrusters designed to stabilize its body during dynamic walking over these narrow paths. The project involves modeling the robot using the HROM (Husky Reduced Order Model) and developing an optimal control framework. This framework is based on polynomial approximation of the HROM and a collocation approach to derive optimal thruster commands necessary for achieving dynamic walking on narrow paths. The effectiveness of the modeling and control design approach is validated through simulations conducted using Matlab.

Figures (9)

• Figure 1: The Northeastern University Husky Carbon walking on a tube.
• Figure 2: Illustrates the Husky full-fidelity model vs. the reduced order model with the model parameters used in the derivations in Section \ref{['sec:hrom']}.
• Figure 3: The proposed control system flowchart. The user inputs include forward velocity reference and gait parameters.
• Figure 4: Illustrates the stick-diagram of HROM traversing a narrow path with 3D CoM and leg-end trajectories
• Figure 5: Illustrates body position and orientation $\bm q$ and $\bm{\dot{q}}$ under $\bm u$ and governing dynamics given by Eq. \ref{['eq:manipulator eq']}