Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Automated remote industrial inspection platform using spot
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_ | 1867614134866018304 |
|---|---|
| access_status_str | Open Access |
| author | Roux, Dominic |
| author2 | Booysen, Thinus |
| author_browse | Booysen, Thinus Roux, Dominic |
| author_facet | Booysen, Thinus Roux, Dominic |
| author_sort | Roux, Dominic |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Thesis (MEng)--Stellenbosch University, 2023. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/127385 |
| institution | Stellenbosch University (South Africa) |
| language | en_ZA en_ZA |
| last_indexed | 2026-06-10T12:47:13.037Z |
| 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:10019.1/127385 Automated remote industrial inspection platform using spot Roux, Dominic Booysen, Thinus Stellenbosch University. Faculty of Engineering. Dept. of Electrical and Electronic Engineering. Robots, Industrial Human-robot interaction Industry 4.0 Machine learning Thesis (MEng)--Stellenbosch University, 2023. ENGLISH ABSTRACT: In an industrial environment, good housekeeping practices are essential to ensuring safety and efficiency on site. In this context, housekeeping refers to keeping an industrial site clean of loose equipment and debris, which present slip, trip and fall hazards. Managing these hazards often falls upon highly skilled managerial staff, with numerous other urgent responsibilities. As a result, the laborious and time-consuming discipline of regular housekeeping inspections is easily neglected. Enabled by advancements in mobile robotics, autonomous legged robots are increasingly applied to the automation of industrial inspections. However, the automation of on-site housekeeping is an unexplored area in this field of research. In this project, an automated remote inspection platform is developed for the novel application of automating housekeeping inspections in a mining processing plant. It uses a Boston Dynamics Spot for autonomous data collection along a pre-determined inspection route. To enable data collection and transfer from the robot, a custom Raspberry Pi payload is developed that facilitates data transfer from the robot to a GPU-enabled computation endpoint. The remote inspection platform implements a vision-based hazard detection and reporting system, based on a Mask R-CNN computer vision model. While real-time performance is desirable, it is not possible with cloud computing in lieu of dedicated on-site computation. Experimental analysis revealed a 0.442 FPS lower bound on the system performance with cloud computing, due to a significant network overhead. The system uses a novel hazard risk estimation algorithm to classify detected hazards as high or low risk. It evaluates hazard detection and walkway segmentation masks generated by the Mask R-CNN model. The masks are used to determine the relative placement of the hazards on the walkway, by which the risk is estimated. The final system was shown to classify hazard risks accurately 93.22% of the time. The available input sensors with Spot were considered, namely the robot’s built-in stereo grayscale cameras, the latter with an additional depth channel, and the Spot CAM+IR fisheye cameras. Three separate image data sets were collected on site at the Rosh Pinah zinc and lead mine for this evaluation. It was found that the Spot CAM+IR fisheye camera results in the best performing system, achieving a Mask R-CNN precision and recall of 84.4% and 76.4%, respectively. The final system was shown to have an accuracy of 58.79% when trained on a very small data set. The system accuracy could easily be increased to 89.53% and above by significantly increasing the amount of training data. The accuracy could even be increased to meet industrial safety standards, thereby making the system feasible for real-life operations. AFRIKAANS ABSTRACT: In ’n industri¨ele omgewing is die praktyk van goeie huishouding noodsaaklik om veiligheid en doeltreffendheid op die perseel te verseker. Hier verwys huishouding na die perseel skoonhou van los toerusting en puin wat gly, struikel en val gevare inhou. Die bestuur van hierdie gevare kom dikwels op hoogs vaardige bestuurspersoneel met talle ander dringende verantwoordelikhede neer. As gevolg hiervan word die moeisame en tydrowende dissipline van gereelde huishouding inspeksies dikwels verwaarloos. Geaktiveer deur die vooruitgang van mobiele robotika word outonome, lopende robotte toenemend op die outomatisering van industri¨ele inspeksies toegepas. Die outomatisering van huishouding op die perseel is egter ’n onontginde area van hierdie navorsingsveld. In hierdie projek word ’n outomatiese, verafgele¨e inspeksie platform ontwikkel vir die nuwe toepassing van huishouding inspeksies op ’n mynverwerkingsaanleg outomatiseer. Dit gebruik ’n Boston Dynamics Spot om outonome data-insameling langs ’n voorafbepaalde inspeksie roete te doen. Om data-insameling en oordrag te bewerkstellig, word ’n pasgemaakte Raspberry Pi loonvrag ontwikkel om data-oordrag van die robot na ’n videokaart-toegeruste verwerkerseenheid te fasiliteer. Die verafgele¨e inspeksie platform implementeer ’n visie-gebaseerde gevaaropsporing-en-verslaggewingstelsel, gebaseer op ’n Mask R-CNN rekenaarvisie model. Al is intydse prestasie wenslik, dit is nie moontlik met wolkverwerking in plaas van toegewyde verwerking op die perseel nie. Deur steekproefanalise is ’n ondergrens van 0.442 rame per sekonde op die stelselprestasie met wolkverwerking uitgelig, as gevolg van ’n merkwaardige netwerk bokoste. Die stelsel gebruik ’n nuwe gevaar risiko beraming algoritme om opgespoorde gevare as ho¨e of lae risiko te klassifiseer. Dit evalueer die gevaaropsporing- en angpadsegmenteringsmaskers wat deur die Mask R-CNN model gegenereer is. Die maskers is gebruik om die relatiewe plasing van gevare op die gangpad te bepaal, waarvolgens die risiko beraam word. Die finale stelsel het risikos 93.22% van die kere akkuraat geklassifiseer. Die beskikbare toevoersensors van Spot is oorweeg, naamlik die robot se ingeboude stereo grysskaal kameras, die laasgenoemde met ’n bykomende dieptekanaal en die Spot CAM+IR visoogkameras. Drie aparte beelddatastelle is op perseel by die Rosh Pinah sink- en loodmyn vir hierdie evaluasie ingesamel. Die Spot CAM+IR visoogkamera het die beste stelselprestasie gelewer, met ’n Mask R-CNN presisie en herroep van 84.4% en 76.4%, onderskeidelik. Die finale stelsel het ’n akkuraatheid van 58.79% behaal met ’n baie klein opleidingsdatastel. Die stelselakkuraatheid kan maklik tot 89.53% en meer verbeter word deur die aantal opleidingsdata te vermeerder. Die akkuraatheid kan selfs genoeg verbeter word om industrie¨ele veiligheidstandaarde te bereik en daardeur die stelsel haalbaar te maak vir werklike bedrywighede. Masters 2023-03-03T09:16:53Z 2023-05-18T07:19:27Z 2023-03-03T09:16:53Z 2023-05-18T07:19:27Z 2023-03 Thesis http://hdl.handle.net/10019.1/127385 en_ZA en_ZA Stellenbosch University xv, 100 pages : illustrations application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | Robots, Industrial Human-robot interaction Industry 4.0 Machine learning Roux, Dominic Automated remote industrial inspection platform using spot |
| title | Automated remote industrial inspection platform using spot |
| title_full | Automated remote industrial inspection platform using spot |
| title_fullStr | Automated remote industrial inspection platform using spot |
| title_full_unstemmed | Automated remote industrial inspection platform using spot |
| title_short | Automated remote industrial inspection platform using spot |
| title_sort | automated remote industrial inspection platform using spot |
| topic | Robots, Industrial Human-robot interaction Industry 4.0 Machine learning |
| url | http://hdl.handle.net/10019.1/127385 |
| work_keys_str_mv | AT rouxdominic automatedremoteindustrialinspectionplatformusingspot |