Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Trajectory optimisation inspired controller design of a low-cost bipedal robot
Thesis (MEng)--Stellenbosch University, 2023.
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | en_ZA en_ZA |
| Published: |
Stellenbosch : Stellenbosch University
2023
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613794732081152 |
|---|---|
| access_status_str | Open Access |
| author | Erasmus, Devlon |
| author2 | Fisher, Callen |
| author_browse | Erasmus, Devlon Fisher, Callen |
| author_facet | Fisher, Callen Erasmus, Devlon |
| author_sort | Erasmus, Devlon |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Thesis (MEng)--Stellenbosch University, 2023. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/127370 |
| institution | Stellenbosch University (South Africa) |
| language | en_ZA en_ZA |
| last_indexed | 2026-06-10T12:41:46.856Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2023 |
| publishDateRange | 2023 |
| publishDateSort | 2023 |
| publisher | Stellenbosch : Stellenbosch University |
| publisherStr | Stellenbosch : Stellenbosch University |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/127370 Trajectory optimisation inspired controller design of a low-cost bipedal robot Erasmus, Devlon Fisher, Callen Stellenbosch University. Faculty of Engineering. Dept. of Electrical and Electronic Engineering. Bipedalism Trajectory optimization Robotics Thesis (MEng)--Stellenbosch University, 2023. ENGLISH ABSTRACT: For robotic systems to become a greater part of everyday life, low-cost platforms are thus needed with the same capabilities as their expensive counterparts. Legged robotics has become the focal point of recent years, partially aided by the developments in trajectory optimisation. However, a major problem in legged robots is that the precise determinants of locomotion are still under debate, making it difficult to design controllers for these systems. The work done in this research investigated if trajectory optimisation can be used to inspire controller design for a low-cost bipedal robot. A low-cost robotic platform was designed and built to test and validate the trajectories. The robot consisted of servo motors at the hips and pneumatic cylinders as legs and required a support rig to constrain it to the sagittal plane. ROS was used to communicate between the systems running on the robot and the main control loop running on a PC. Trajectory optimisation was used to generate acceleration, steady-state and deceleration trajectories for the robot. The trajectories were based on the mathematical model of the robot, with the contacts modelled using through-contact methods. This method allowed the optimiser to choose when the contacts occurred, which allowed for an optimal gait to emerge from the trajectories. Once the robot was finalised and built, the trajectories generated by the trajectory optimisation were used to inspire the design of controllers for the system by analysing common trends within the trajectories. It was found that during the steady-state phase of the robot, the Raibert controller emerged, resulting in the controller taking a state-machine form similar to the one proposed by Raibert. The controllers were first tested in a ROS Gazebo simulation as a validation step to investigate if the controllers could successfully control the system before implementing it on the physical robot. The controllers were implemented on a fixed and free body model and were tested on a rigid surface as well as a rough surface, in order to test their robustness. The results of the testing indicated that trajectory optimisation could lead to controller design for a low-cost biped robot. AFRIKAANS OPSOMMING: Om ten einde robotstelsels ’n groter deel van die alledaagse lewe te maak, is laekosteplatforms nodig met dieselfde vermoëns as hul duur eweknieë. Been-robotika het die fokuspunt van onlangse jare geword, aangehelp deur die ontwikkelings in trajek-optimering. ’n Groot probleem in robote met bene is egter dat die presiese determinante van voortbeweging steeds onder debat is, wat dit moeilik maak om beheerders vir hierdie stelsels te ontwerp. Die werk wat in hierdie navorsing gedoen was, ondersoek of trajek-optimering ’n beheerstelser ontwerp vir ’n laekoste tweevoetige robot kan inspireer. Vir daai rede was ’n laekoste robotplatform ontwerp en gebou om die trajekte op te toets en te valideer. Die robot bestaan uit ’n servomotors by die heupe en pneumatiese silinders as bene en het ’n ondersteuningstuig nodig om dit tot die sagittale vlak te beperk. ROS was gebruik om te kommunikeer tussen die stelsels wat op die robot en die hoofbeheerlus wat op ’n rekenaar loop. Trajek-optimering was gebruik om versnelling, bestendige-toestand en vertraging trajekte vir die robot te genereer. Die bane was gebaseer op die wiskundige model van die robot, met die kontakte wat deur “through-contact” metodes gemodelleer was. Hierdie metode het die optimaliseerder toegelaat om te kies wanneer die kontakte plaasvind, wait toegelaat het dat ’n optimale loop gang uit die trajekte na vore kom. Sodra die robot gefinaliseer en gebou was, was die trajekte wat deur die optimalisering gegenereer was, gebruik om beheerders vir die stelsel te ontwerp deur algemene neigings wat in die trajek na vore kom te ontleed. Daar was gevind dat tydens die bestendige fase van die robot die Raibert-beheerder na vore gekom het. Die beheerder het dus ’n staatsmasjien-vorm aangeneem soortgelyk aan die een wat deur Raibert voorgestel is. Die beheerders was eers in ’n ROS Gazebo-simulasie getoets as ’n valideringsstap om te ondersoek of die beheerders ’n stelsel suksesvol kan beheer voordat dit op die fisiese robot geïmplementeer word. Die beheerders was op ’n vaste en vrye liggaamsmodel geïmplementeer en was op ’n rigiede oppervlak sowel as ’n growwe oppervlak getoets om die robuustheid van die beheerders te ondersoek. Die resultate van die toetsing het aangedui dat baanoptimering kan lei tot beheerderontwerp vir ’n laekoste tweevoetige robot. Masters 2023-03-03T09:11:42Z 2023-05-18T07:18:37Z 2023-03-03T09:11:42Z 2023-05-18T07:18:37Z 2023-03 Thesis http://hdl.handle.net/10019.1/127370 en_ZA en_ZA Stellenbosch University xii, 92 pages : illustrations application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Bipedalism Trajectory optimization Robotics Erasmus, Devlon Trajectory optimisation inspired controller design of a low-cost bipedal robot |
| title | Trajectory optimisation inspired controller design of a low-cost bipedal robot |
| title_full | Trajectory optimisation inspired controller design of a low-cost bipedal robot |
| title_fullStr | Trajectory optimisation inspired controller design of a low-cost bipedal robot |
| title_full_unstemmed | Trajectory optimisation inspired controller design of a low-cost bipedal robot |
| title_short | Trajectory optimisation inspired controller design of a low-cost bipedal robot |
| title_sort | trajectory optimisation inspired controller design of a low cost bipedal robot |
| topic | Bipedalism Trajectory optimization Robotics |
| url | http://hdl.handle.net/10019.1/127370 |
| work_keys_str_mv | AT erasmusdevlon trajectoryoptimisationinspiredcontrollerdesignofalowcostbipedalrobot |