Joe Payne

j.joseph.payne [at] gmail.com

My work focuses on state estimation for legged robots; I'm specifically interested in what happens during contact events such as foot touchdown during locomotion. The impulsive dynamics during these events create nonsmooth and potentially discontinuous dynamics which can cause many standard algorithms for control and state estimation to underperform or simply fail. In my PhD I have sought to create control and state estimation methods that are able to explicitly handle these contact events. I work on these problems with Aaron Johnson in the Robomechanics Lab at Carnegie Mellon University. I expect to complete my PhD in spring 2024 and I am looking for employment opportunities for summer or fall 2024. My resume is available here.

Research (And Projects)

Filtering on Hybrid Systems

This project focuses on the mathematical properties of hybrid systems with state driven mode transitions (such as contact dynamics in legged robots). The goal is to understand how uncertainty evolves in hybrid systems to better estimate current state. The first paper on this topic was published in Automatica, a preprint version can also be found on arXiv. The sequel paper which focuses on interacting with uncertain guards appeared at IROS 2022 and a preprint can be found on arXiv.

SLAM on Legged Robots

Many modern SLAM systems provide inaccurate estimates of motion on legged robots, if they can provide estimates at all. This project is interested in determining what it is about the motion that causes failure, then using that information to improve SLAM algorithms and optimize gaits for state estimation. The paper was presented at ICRA 2022 as Periodic SLAM: Using Cyclic Constraints to Improve the Performance of Visual-Inertial SLAM on Legged Robots.

FlamingoBot

This was a robot created for the class Robot Design and Experiementation. The idea behind the robot is that many current legged robots consume a great deal of energy just to remain in a stationary upright position. In response to this, we took inspiration from the flamingo to reduce the energy to remain standing to zero while on a flat surface. We also made an overly dramatic video on the robot.

Talks

State Estimation for Hybrid Systems with Uncertain Guards

This talk discusses the extention of saltation matrices to the consideration of uncertainty in the system model. This includes uncertainty in the guard location, which could take the form of unknown ground height for a legged robot, and uncertainty in reset map parameters, such as an unknown coefficient of restitution in an impact system. This talk was presented at IROS 2022.

Saltation Matrices and Why You Should Care

This talk covers the fundamentals of the saltation matrix, including a brief geometric derivation and how it can be applied when adapting smooth algorithms to hybrid dynamical systems. This was presented at Dynamic Walking 2022.

FlamingoBot: A Zero Energy Standing Bipedal Robot

This talk, jointly presented with my collaborator Nathan Kong, introduces FlamingoBot, a robot which we designed to be able to remain in an upright standing position with no required motor torques. This design was created to save energy in robots which spend large portions of their operating times standing still. This was presented at Dynamic Walking 2019.

Teaching

Dynamic Systems and Controls

Mechanical Engineering 352

The second class in dynamics for MechE undergrads. The main focus is on frequency domain analysis - transfer functions, bode plots, root loci, etc. It briefly introduces the concept of feedback control.

Winter 2021 · TA

Dynamics

Mechanical Engineering 351

The first class in dynamics for MechE undergrads. It deals primarily with force and energy methods for creating equations of motion for dynamic systems.

Fall 2019 · TA

Interests

Outside of academics, you can find me rock climbing all over Pennsylvania, West Virginia, and Ohio. I'm primarily a boulderer, but can occasionally be convinced to do some sport climbing. I am also a big fan of basically any board sport: snowboarding, skateboarding, surfing, wakeboarding, etc!

For less action-heavy activities, I try to do a lot of reading. I primarily read literary fiction, but occasionally read essay collections or spy novels. Right now, I can't recommend The Overstory, American Spy, and Go Ahead in the Rain enough!