kinematics

This work extends my own contributions at the tail end of my masters thesis with some excellent novel ideas by Yanhao Yang at Oregon State University. We perform an optimization of body orientation frame for mobile systems, key for motion planning using models for dynamics like constraint curvature. However, we address the curse of dimensionality by performing optimization in the region surrounding the current motion plan, rather than for the entire configuration space.

This is the work of an REU student of mine, Bill Fan, which he started at Oregon State within the Laboratory for Robotics and Applied Mechanics and continued on his own at Olin College of Engineering. It was presented at RoboSoft 2023. We use the Lie algebra to produce an estimate of manipulator configuration for soft, constant-twist manipulators.

In this work, presented at IROS 2022, we generate families of optimal motion plans for biomimetic systems. These families are parametrized by a single variable, providing an intuitive control mode for systems with very complex dynamics.

This is my masters thesis, the culmination of my geometric mechanics work at Oregon State University. I explore models for locomotion based on the Lie algebra, consequences of body frame coordinate optimization, and dimensionality reduction for biomimetic mobile systems.

As part of a project with two other robotics students, I explored kinematic path planning for a manipulator with uncertainty in the actuators. This amounted to generating an uncertainty metric (the covariance of the actuators) in the joint space, which was then projected into the task space to act as a heuristic for path planning.

In this work, presented at ICRA 2022, we demonstrate the role of body-frame coordinate optimization in reducing error when using simplified models for system dynamics. This was my first conference paper.

This project applies Lagrangian dynamics and some of the structures in geometric mechanics to model and control the dynamics of a manipulator that has series elastic actuators. The purpose of this project was to explore the idea of traditional manipulators with tunable compliance, which could permit working in both industrial applications and alongside people.

As part of an extended project, we were interested in producing highly dynamic legged locomotion (think running, jumping) on Agility Robotics’ Cassie. My role in this project built on some previous work by my advisor, Prof. Ross L. Hatton, to optimally choose stable body frame coordinates for this robot.

As part of a graduate course and in support of my own research, I explored methods of visualizing robot configuration spaces. This is particularly useful as these spaces grow in size, providing intuition that would otherwise be lost. This work was later expanded in my own thesis and supported other publications.