Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Real-time occupancy grid mapping using LSD-SLAM
Thesis (MScEng)--Stellenbosch University, 2017.
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | en_ZA |
| Published: |
Stellenbosch : Stellenbosch University
2017
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867613774408581120 |
|---|---|
| access_status_str | Open Access |
| author | Hull, Graham |
| author2 | Van Daalen, C. E. |
| author_browse | Hull, Graham Van Daalen, C. E. |
| author_facet | Van Daalen, C. E. Hull, Graham |
| author_sort | Hull, Graham |
| collection | Thesis |
| dc_rights_str_mv | Stellenbosch University |
| description | Thesis (MScEng)--Stellenbosch University, 2017. |
| format | Thesis |
| id | oai:scholar.sun.ac.za:10019.1/102780 |
| institution | Stellenbosch University (South Africa) |
| language | en_ZA |
| last_indexed | 2026-06-10T12:41:29.531Z |
| license_str | Other — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from SUNScholar — Stellenbosch University Repository |
| publishDate | 2017 |
| publishDateRange | 2017 |
| publishDateSort | 2017 |
| publisher | Stellenbosch : Stellenbosch University |
| publisherStr | Stellenbosch : Stellenbosch University |
| record_format | dspace |
| source_str | SUNScholar — Stellenbosch University Repository |
| spelling | oai:scholar.sun.ac.za:10019.1/102780 Real-time occupancy grid mapping using LSD-SLAM Hull, Graham Van Daalen, C. E. Stellenbosch University. Faculty of Engineering. Dept. of Electrical and Electronic Engineering. LSD-SLAM UCTD Grid plans (City planning) Real-time data processing -- Navigation Grids (Cartography) Thesis (MScEng)--Stellenbosch University, 2017. ENGLISH ABSTRACT: This thesis investigates the use of semi-dense depth data from monocular vision using large scale direct SLAM (LSD-SLAM) to create accurate occupancy grid maps for autonomous navigation, in real-time. Having an accurate map is crucial for an autonomous system to avoid collisions and remain safe within its environment. Sensors used to gather information on the environment are typically associated with some degree of uncertainty, and this must be considered when building a map. An autonomously navigating system also needs to have clear definition of free and occupied space within its environment. Literature shows that LSD-SLAM has great potential as a highly accurate and real-time SLAM algorithm; however, the resulting map is in the form of a semi-dense point-cloud which is not immediately useful to an autonomously navigating system. The point-cloud map must therefore be processed further. Occupancy grid maps (OGMs) offer an ideal solution for map representation that is useful for autonomous navigation. The environment is divided into evenly spaced grid cells, each representing a probability of occupancy. OGMs also allow maps to be efficiently updated with the incorporation of uncertainty from sensor measurements. Inverse sensor models (ISMs) can be used to characterise the uncertainty of a particular sensor and to calculate the prediction of occupancy given a sensor measurement and its uncertainty. A literature review shows that two popular ISMs (one by Thrun and one by Andert) can be used in conjunction with the depth estimates of LSD-SLAM to create an OGM. Literature also shows that each of these ISMs contains a parameter that is associated with very little information on how their values should be chosen, and we therefore included this in our investigation. We design a mapping system using the aforementioned ISMs, which runs in parallel with LSD-SLAM. Initial tests show that the performance of the open-source version of LSD-SLAM did not agree with the author’s claims. The results also revealed a significant lack of sufficient datasets for our main evaluations on map accuracy. Our mapping system was tested on 3 main criteria: memory usage, performance and accuracy. All evaluations were performed on both ISMs, on various datasets, over a range resolutions and parameter changes for each ISM. Results showed that Thrun’s ISM out-performed Andert’s ISM on all criteria, and that our system could indeed produce accurate maps that could be useful for autonomous navigation. The results also showed that the “default” choice for the parameters of each ISM is not necessarily always sufficient. Additionally, we conclude that LSD-SLAM does not perform well in terms of 30 Hz real-time requirements, while our mapping system can. AFRIKAANSE OPSOMMING: Hierdie tesis ondersoek die gebruik van semi-digte diepte data van monovisie deur gebruik te maak van groot skaal direkte SLAM (LSD-SLAM) om akkurate besetting-rooster-kaarte vir outonome navigasie intyds op te stel. Om ’n akkurate kaart te hê is krities vir ’n outonoome stelsel om botsings te vermy en om veilig te bly in die omgewing. Sensors wat gebruik word om inligting oor die omgewing in te samel, word tipies geassosieer met ’n mate van onsekerheid, en dit moet in ag geneem word wanneer ’n kaart gebou word. ’n Outonome navigasiestelsel moet ook duidelike definisie van vrye en besette ruimte binne sy omgewing hê. Literatuur wys dat LSD-SLAM groot potensiaal het as ’n hoogs akkurate en intydse SLAM algoritme, maar die resulterende kaart is egter verteenwoordig as ’n semi-digte punt-wolk, wat nie onmiddellik bruikbaar is vir outonome navigasie stelsels nie. Die punt-wolkkaart moet dus verder verwerk word. Besetting-rooster-kaarte (OGMs) bied ’n ideale oplossing vir kaartvoorstelling wat nuttig is vir outonome navigasie. Die omgewing word verdeel in eweredig verspreide roosterselle, wat elkeen ’n waarskynlikheid van besetting verteenwoordig. OGMs laat ook toe om kaarte doeltreffend op te dateer met die insluiting van onsekerheid van sensormetings. Inverse sensor modelle (ISM’s) kan gebruik word om die onsekerheid van ’n spesifieke sensor te karakteriseer en die voorspelling van besetting te bereken wat ’n sensormeting en sy onsekerheid gegee het. ’n Literatuuroorsig toon dat twee gewilde ISM’s (een deur Thrun en een deur Andert) saam met die diepte ramings van LSD-SLAM gebruik kan word om ’n OGM te skep. Literatuur toon ook dat elkeen van hierdie ISM’s ’n parameter bevat wat geassosieer word met baie min inligting oor hoe hul waardes gekies moet word, en ons het dit dus by ons ondersoek ingesluit. Ons ontwerp ’n kartering-stelsel met behulp van die voorafgemelde ISMs, wat in parallel met LSD-SLAM uitvoer. Aanvanklike toetse toon dat die uitvoering van die oopbron weergawe van LSD-SLAM nie met die skrywer se eise ooreenstem nie. Die resultate het ook ’n aansienlike gebrek aan voldoende datastelle vir ons hoofevaluasies op kaart akkuraatheid geopenbaar. Ons kartering-stelsel is getoets op 3 hoofkriteria: geheueverbruik, spoed en akkuraatheid. Alle evaluasies is uitgevoer op beide ISM’s, op verskeie datastelle, oor ’n reeks resolusies en parameter veranderinge vir elke ISM. Resultate het getoon dat Thrun se ISM Andert se ISM op alle kriteria uitpresteer het, en dat ons stelsel inderdaad akkurate kaarte kan produseer wat nuttig kan wees vir outonome navigasie. Die resultate het ook getoon dat die “standaard” keuse vir die parameters van elke ISM nie noodwendig altyd voldoende is nie. Daarbenewens kom ons tot die gevolgtrekking dat LSD-SLAM nie goed presteer in terme van 30 Hz intyds vereistes, terwyl ons kartering-stelsel kan. 2017-11-20T07:41:16Z 2017-12-11T10:53:28Z 2017-11-20T07:41:16Z 2017-12-11T10:53:28Z 2017-12 Thesis http://hdl.handle.net/10019.1/102780 en_ZA Stellenbosch University 128 pages : illustrations application/pdf Stellenbosch : Stellenbosch University |
| spellingShingle | LSD-SLAM UCTD Grid plans (City planning) Real-time data processing -- Navigation Grids (Cartography) Hull, Graham Real-time occupancy grid mapping using LSD-SLAM |
| title | Real-time occupancy grid mapping using LSD-SLAM |
| title_full | Real-time occupancy grid mapping using LSD-SLAM |
| title_fullStr | Real-time occupancy grid mapping using LSD-SLAM |
| title_full_unstemmed | Real-time occupancy grid mapping using LSD-SLAM |
| title_short | Real-time occupancy grid mapping using LSD-SLAM |
| title_sort | real time occupancy grid mapping using lsd slam |
| topic | LSD-SLAM UCTD Grid plans (City planning) Real-time data processing -- Navigation Grids (Cartography) |
| url | http://hdl.handle.net/10019.1/102780 |
| work_keys_str_mv | AT hullgraham realtimeoccupancygridmappingusinglsdslam |