Änderungen
Auf 24. November 2023 um 13:33:02 UTC, Felix Fröhling:
-
Die neue Ressource Article wurde zu Generation of Correction Data for Autonomous Driving by Means of Machine Learning and On-Board Diagnostics hinzugefügt
f | 1 | { | f | 1 | { |
2 | "author": "[{\"author\": \"Felix Fr\u00f6hling\", \"author_email\": | 2 | "author": "[{\"author\": \"Felix Fr\u00f6hling\", \"author_email\": | ||
3 | \"Felix.Froehling@carissma.eu\"}]", | 3 | \"Felix.Froehling@carissma.eu\"}]", | ||
4 | "author_email": null, | 4 | "author_email": null, | ||
5 | "creator_user_id": "fd88527d-4fb2-4f5a-8ca5-de730b63d31c", | 5 | "creator_user_id": "fd88527d-4fb2-4f5a-8ca5-de730b63d31c", | ||
6 | "end_collection_date": "", | 6 | "end_collection_date": "", | ||
7 | "groups": [ | 7 | "groups": [ | ||
8 | { | 8 | { | ||
9 | "description": "", | 9 | "description": "", | ||
10 | "display_name": "Datensatz und Dokumente", | 10 | "display_name": "Datensatz und Dokumente", | ||
11 | "id": "a6939c14-5825-45af-b335-82712b5db05f", | 11 | "id": "a6939c14-5825-45af-b335-82712b5db05f", | ||
12 | "image_display_url": | 12 | "image_display_url": | ||
13 | "https://catalog.savenow.de/base/images/group_icons/dataset.jpg", | 13 | "https://catalog.savenow.de/base/images/group_icons/dataset.jpg", | ||
14 | "name": "dataset", | 14 | "name": "dataset", | ||
15 | "title": "Datensatz und Dokumente" | 15 | "title": "Datensatz und Dokumente" | ||
16 | }, | 16 | }, | ||
17 | { | 17 | { | ||
18 | "description": "", | 18 | "description": "", | ||
19 | "display_name": "Mobilit\u00e4t", | 19 | "display_name": "Mobilit\u00e4t", | ||
20 | "id": "fb9ec11d-93e2-4708-88a1-25019bb954b5", | 20 | "id": "fb9ec11d-93e2-4708-88a1-25019bb954b5", | ||
21 | "image_display_url": | 21 | "image_display_url": | ||
22 | "https://catalog.savenow.de/base/images/group_icons/mobility.svg", | 22 | "https://catalog.savenow.de/base/images/group_icons/mobility.svg", | ||
23 | "name": "mobility", | 23 | "name": "mobility", | ||
24 | "title": "Mobilit\u00e4t" | 24 | "title": "Mobilit\u00e4t" | ||
25 | } | 25 | } | ||
26 | ], | 26 | ], | ||
27 | "id": "becd784d-1096-4893-a102-a9d496ef8ae2", | 27 | "id": "becd784d-1096-4893-a102-a9d496ef8ae2", | ||
28 | "isopen": true, | 28 | "isopen": true, | ||
29 | "language": "en", | 29 | "language": "en", | ||
30 | "licence_agreement": [ | 30 | "licence_agreement": [ | ||
31 | "licence_agreement_check" | 31 | "licence_agreement_check" | ||
32 | ], | 32 | ], | ||
33 | "license_id": "CC0-1.0", | 33 | "license_id": "CC0-1.0", | ||
34 | "license_title": "Creative Commons Zero (CC0 1.0)", | 34 | "license_title": "Creative Commons Zero (CC0 1.0)", | ||
35 | "license_url": "https://creativecommons.org/publicdomain/zero/1.0/", | 35 | "license_url": "https://creativecommons.org/publicdomain/zero/1.0/", | ||
36 | "maintainer": "[{\"Maintainer Email\": | 36 | "maintainer": "[{\"Maintainer Email\": | ||
37 | \"eduardo.sanchezmorales@thi.de\", \"maintainer\": \"Alberto Flores | 37 | \"eduardo.sanchezmorales@thi.de\", \"maintainer\": \"Alberto Flores | ||
38 | Fern\u00e1ndez\", \"phone\": \"\", \"role\": \"contact person\"}]", | 38 | Fern\u00e1ndez\", \"phone\": \"\", \"role\": \"contact person\"}]", | ||
39 | "maintainer_email": null, | 39 | "maintainer_email": null, | ||
40 | "metadata_created": "2023-11-24T13:31:14.056396", | 40 | "metadata_created": "2023-11-24T13:31:14.056396", | ||
n | 41 | "metadata_modified": "2023-11-24T13:31:14.056402", | n | 41 | "metadata_modified": "2023-11-24T13:33:02.341362", |
42 | "name": "generation-of-correction-data-for-autonomous-driving", | 42 | "name": "generation-of-correction-data-for-autonomous-driving", | ||
43 | "notes": "Abstract: A highly accurate reference vehicle state is a | 43 | "notes": "Abstract: A highly accurate reference vehicle state is a | ||
44 | requisite for the evaluation and validation\r\nof Autonomous Driving | 44 | requisite for the evaluation and validation\r\nof Autonomous Driving | ||
45 | (AD) and Advanced Driver Assistance Systems (ADASs). This | 45 | (AD) and Advanced Driver Assistance Systems (ADASs). This | ||
46 | highly\r\naccurate vehicle state is usually obtained by means of | 46 | highly\r\naccurate vehicle state is usually obtained by means of | ||
47 | Inertial Navigation Systems (INSs) that obtain\r\nposition, velocity, | 47 | Inertial Navigation Systems (INSs) that obtain\r\nposition, velocity, | ||
48 | and Course Over Ground (COG) correction data from Satellite Navigation | 48 | and Course Over Ground (COG) correction data from Satellite Navigation | ||
49 | (SatNav).\r\nHowever, SatNav is not always available, as is the case | 49 | (SatNav).\r\nHowever, SatNav is not always available, as is the case | ||
50 | of roofed places, such as parking structures,\r\ntunnels, or urban | 50 | of roofed places, such as parking structures,\r\ntunnels, or urban | ||
51 | canyons. This leads to a degradation over time of the estimated | 51 | canyons. This leads to a degradation over time of the estimated | ||
52 | vehicle state.\r\nIn the present paper, a methodology is proposed that | 52 | vehicle state.\r\nIn the present paper, a methodology is proposed that | ||
53 | consists on the use of a Machine Learning\r\n(ML)-method (Transformer | 53 | consists on the use of a Machine Learning\r\n(ML)-method (Transformer | ||
54 | Neural Network\u2014TNN) with the objective of generating highly | 54 | Neural Network\u2014TNN) with the objective of generating highly | ||
55 | accurate\r\nvelocity correction data from On-Board Diagnostics (OBD) | 55 | accurate\r\nvelocity correction data from On-Board Diagnostics (OBD) | ||
56 | data. The TNN obtains OBD data as input\r\nand measurements from | 56 | data. The TNN obtains OBD data as input\r\nand measurements from | ||
57 | state-of-the-art reference sensors as a learning target. The results | 57 | state-of-the-art reference sensors as a learning target. The results | ||
58 | show that\r\nthe TNN is able to infer the velocity over ground with a | 58 | show that\r\nthe TNN is able to infer the velocity over ground with a | ||
59 | Mean Absolute Error (MAE) of 0.167 km/h (0.046 m/s) when a database of | 59 | Mean Absolute Error (MAE) of 0.167 km/h (0.046 m/s) when a database of | ||
60 | 3,428,099 OBD measurements is considered. The accuracy decreases to | 60 | 3,428,099 OBD measurements is considered. The accuracy decreases to | ||
61 | 0.863 km/h (0.24 m/s) when only 5000 OBD measurements are used. Given | 61 | 0.863 km/h (0.24 m/s) when only 5000 OBD measurements are used. Given | ||
62 | that the obtained accuracy\r\nclosely resembles that of | 62 | that the obtained accuracy\r\nclosely resembles that of | ||
63 | state-of-the-art reference sensors, it allows INSs to be provided with | 63 | state-of-the-art reference sensors, it allows INSs to be provided with | ||
64 | accurate\r\nvelocity correction data. An inference time of less than | 64 | accurate\r\nvelocity correction data. An inference time of less than | ||
65 | 40 ms for the generation of new correction\r\ndata is achieved, which | 65 | 40 ms for the generation of new correction\r\ndata is achieved, which | ||
66 | suggests the possibility of online implementation. This supports a | 66 | suggests the possibility of online implementation. This supports a | ||
67 | highly\r\naccurate estimation of the vehicle state for the evaluation | 67 | highly\r\naccurate estimation of the vehicle state for the evaluation | ||
68 | and validation of AD and ADAS, even in\r\nSatNav-deprived | 68 | and validation of AD and ADAS, even in\r\nSatNav-deprived | ||
69 | environments.", | 69 | environments.", | ||
n | 70 | "num_resources": 0, | n | 70 | "num_resources": 1, |
71 | "num_tags": 1, | 71 | "num_tags": 1, | ||
72 | "organization": { | 72 | "organization": { | ||
73 | "approval_status": "approved", | 73 | "approval_status": "approved", | ||
74 | "created": "2023-06-27T14:06:24.055358", | 74 | "created": "2023-06-27T14:06:24.055358", | ||
75 | "description": "", | 75 | "description": "", | ||
76 | "id": "518ef23a-8d7b-4829-9b42-7a182a73cc0f", | 76 | "id": "518ef23a-8d7b-4829-9b42-7a182a73cc0f", | ||
77 | "image_url": "2023-06-27-150145.628103THI.jpg", | 77 | "image_url": "2023-06-27-150145.628103THI.jpg", | ||
78 | "is_organization": true, | 78 | "is_organization": true, | ||
79 | "name": "technische-hochschule-ingolstadt-thi", | 79 | "name": "technische-hochschule-ingolstadt-thi", | ||
80 | "state": "active", | 80 | "state": "active", | ||
81 | "title": "Technische Hochschule Ingolstadt (THI)", | 81 | "title": "Technische Hochschule Ingolstadt (THI)", | ||
82 | "type": "organization" | 82 | "type": "organization" | ||
83 | }, | 83 | }, | ||
84 | "owner_org": "518ef23a-8d7b-4829-9b42-7a182a73cc0f", | 84 | "owner_org": "518ef23a-8d7b-4829-9b42-7a182a73cc0f", | ||
85 | "private": false, | 85 | "private": false, | ||
86 | "relationships_as_object": [], | 86 | "relationships_as_object": [], | ||
87 | "relationships_as_subject": [], | 87 | "relationships_as_subject": [], | ||
t | 88 | "resources": [], | t | 88 | "resources": [ |
89 | { | ||||
90 | "allowed_users": "", | ||||
91 | "cache_last_updated": null, | ||||
92 | "cache_url": null, | ||||
93 | "created": "2023-11-24T13:33:02.373196", | ||||
94 | "datastore_active": false, | ||||
95 | "description": "Generation of Correction Data for Autonomous | ||||
96 | Driving by Means of Machine Learning and On-Board Diagnostics", | ||||
97 | "format": "pdf", | ||||
98 | "hash": "", | ||||
99 | "id": "4965f572-c3b3-4f87-9e89-2ff1cd8b707b", | ||||
100 | "last_modified": "2023-11-24T13:33:02.314454", | ||||
101 | "metadata_modified": "2023-11-24T13:33:02.354311", | ||||
102 | "mimetype": "application/pdf", | ||||
103 | "mimetype_inner": null, | ||||
104 | "name": "Article", | ||||
105 | "package_id": "becd784d-1096-4893-a102-a9d496ef8ae2", | ||||
106 | "position": 0, | ||||
107 | "resource_type": null, | ||||
108 | "restricted": "{\"level\": \"public\"}", | ||||
109 | "size": 13548130, | ||||
110 | "state": "active", | ||||
111 | "url": | ||||
112 | 707b/download/generation_of_correction_data_for_autonomous_drivi.pdf", | ||||
113 | "url_type": "upload" | ||||
114 | } | ||||
115 | ], | ||||
89 | "spatial": "", | 116 | "spatial": "", | ||
90 | "state": "draft", | 117 | "state": "draft", | ||
91 | "tags": [ | 118 | "tags": [ | ||
92 | { | 119 | { | ||
93 | "display_name": "Article", | 120 | "display_name": "Article", | ||
94 | "id": "e80a4ad9-1e6b-48d4-93f9-34e9d89bc67b", | 121 | "id": "e80a4ad9-1e6b-48d4-93f9-34e9d89bc67b", | ||
95 | "name": "Article", | 122 | "name": "Article", | ||
96 | "state": "active", | 123 | "state": "active", | ||
97 | "vocabulary_id": null | 124 | "vocabulary_id": null | ||
98 | } | 125 | } | ||
99 | ], | 126 | ], | ||
100 | "title": "Generation of Correction Data for Autonomous Driving by | 127 | "title": "Generation of Correction Data for Autonomous Driving by | ||
101 | Means of Machine Learning and On-Board Diagnostics", | 128 | Means of Machine Learning and On-Board Diagnostics", | ||
102 | "type": "dataset", | 129 | "type": "dataset", | ||
103 | "url": null, | 130 | "url": null, | ||
104 | "version": "" | 131 | "version": "" | ||
105 | } | 132 | } |