First Advisor

Alex Hunt

Term of Graduation

Fall 2025

Date of Publication

12-28-2025

Document Type

Thesis

Degree Name

Master of Science (M.S.) in Mechanical Engineering

Department

Mechanical and Materials Engineering

Language

English

Subjects

Braided Pneumatic Actuator, Central Pattern Generator, SNS-Toolbox, Spiking Neuron, Synthetic Nervous System

Physical Description

1 online resource (viii, 60 pages)

Abstract

Robotic development started billions of years after earth first began biological prototyping. While the evolutionary process in animals is quite different from robotic prototyping, it still remains no surprise that animals remain more agile, more efficient, and more adaptable than even the most advanced engineered robots to date, simply due to the sheer length of their improvement process. While bridging this improvement gap between deficits in robots and the abilities of animals is difficult, a specific subfield of robotics engineering has attempted to do so using biomimetic robotics. At Portland State, the Agile and Adaptive Robotics Lab (AARL) studies these biomimetic robotics. Its main biomimetic robotic platform, Muscle Mutt, is a quadruped actuated by artificial muscles and based on the proportions of a Whippet dog. While Muscle Mutt is structurally complete, it does not have a biologically-inspired control system to generate locomotion. Thus, this work details the electrical and control improvements to the Muscle Mutt and its test platform with the end goal of generating closed-loop, biologically-inspired air walking.

This work describes the implementation of a 3D printed adapter to fix Muscle Mutt’s non-functioning test treadmill platform. It also describes a force generation method in artificial muscles, called braided pneumatic actuators (BPAs), using neuron spike activations. It also details a novel method for modulating the mean force activation curve in BPAs using a noisy reset neuron model.

The work then chronicles the development of a virtual model of Muscle Mutt in MuJoCo, and the adaptation of a synthetic nervous system from a rat hind limb walking model in SNS-Toolbox to control the virtual model. Using spiking motoneurons to create activations in the virtual muscles, closed-loop, coordinated stepping in the fore and hind limbs is achieved in the simulated model of Muscle Mutt.

The Muscle Mutt robot is then connected to the synthetic nervous system, which was developed using the virtual model of Muscle Mutt. Adaptations of the network simulation for implementation in the physical robot are described, such as how sensory feedback from potentiometers and BPA pressure sensors is processed as biological muscle data for input in the synthetic nervous system (SNS). This biologically-inspired control scheme is able to generate rough, closed-loop stepping in the robot, suggesting the further utility of using spike activations from the synthetic nervous system to generate Muscle Mutt's locomotion.

Rights

In Copyright. URI: http://rightsstatements.org/vocab/InC/1.0/ This Item is protected by copyright and/or related rights. You are free to use this Item in any way that is permitted by the copyright and related rights legislation that applies to your use. For other uses you need to obtain permission from the rights-holder(s).

Comments

This work is supported in part by NSF DBI 2015317 as part of the NSF/CIHR/ DFGFRQ /UKRI-MRCNext Generation Networks for Neuroscience Program.

Persistent Identifier

https://archives.pdx.edu/ds/psu/44406

Download

Included in

Biomechanics Commons

Share

COinS