{
  "id": 185244,
  "title": "What are the definitions of the different coordinate systems?",
  "url": "/competitions/lyft-motion-prediction-autonomous-vehicles/discussion/185244",
  "author_name": "",
  "post_date": "2020-09-20T05:02:22.581546900Z",
  "votes": 1,
  "comment_count": 4,
  "views": 0,
  "content": "<p>I haven't been able to find definitions for the different coordinate systems used in the l5kit library. What are the ecef, world, image_space, and aerial coordinate systems? what are the origins of each? units of each? and directions of the axes of each?  Also, what is the relationships between these coordinate systems?</p>",
  "messages": [
    {
      "id": "1018922",
      "postDate": "09/20/2020 05:02:22",
      "content": "<p>I haven't been able to find definitions for the different coordinate systems used in the l5kit library. What are the ecef, world, image_space, and aerial coordinate systems? what are the origins of each? units of each? and directions of the axes of each?  Also, what is the relationships between these coordinate systems?</p>",
      "rawMarkdown": "I haven't been able to find definitions for the different coordinate systems used in the l5kit library. What are the ecef, world, image_space, and aerial coordinate systems? what are the origins of each? units of each? and directions of the axes of each?  Also, what is the relationships between these coordinate systems?",
      "votes": null
    },
    {
      "id": "1019059",
      "postDate": "09/20/2020 07:03:25",
      "content": "<p>I have gathered that ecef?/enu is used for the protobuf data (lanes, traffic lights), world is the zarr format (ego, agents) and image space is kind of user defined </p>\n<p>There are a couple of hard-coded transform matrices here: <a href=\"https://github.com/lyft/l5kit/blob/master/l5kit/l5kit/rasterization/rasterizer_builder.py\" target=\"_blank\">https://github.com/lyft/l5kit/blob/master/l5kit/l5kit/rasterization/rasterizer_builder.py</a>  see world_to_ecef and ecef_to_aerial</p>\n<p>A bit more info on ecef<br>\n<a href=\"https://www.oc.nps.edu/oc2902w/coord/coord.pdf\" target=\"_blank\">https://www.oc.nps.edu/oc2902w/coord/coord.pdf</a></p>\n<p>Isn't it a 3d surface coordinate system? (both x and y are longitude/ z is lat?)</p>",
      "rawMarkdown": "I have gathered that ecef?/enu is used for the protobuf data (lanes, traffic lights), world is the zarr format (ego, agents) and image space is kind of user defined \n\nThere are a couple of hard-coded transform matrices here: https://github.com/lyft/l5kit/blob/master/l5kit/l5kit/rasterization/rasterizer_builder.py  see world_to_ecef and ecef_to_aerial\n\nA bit more info on ecef\nhttps://www.oc.nps.edu/oc2902w/coord/coord.pdf\n\nIsn't it a 3d surface coordinate system? (both x and y are longitude/ z is lat?)",
      "votes": null
    },
    {
      "id": "1020627",
      "postDate": "09/21/2020 10:06:07",
      "content": "<p>This is probably the best resource we have: <a href=\"https://github.com/lyft/l5kit/blob/master/coords_systems.md\" target=\"_blank\">https://github.com/lyft/l5kit/blob/master/coords_systems.md</a></p>",
      "rawMarkdown": "This is probably the best resource we have: https://github.com/lyft/l5kit/blob/master/coords_systems.md",
      "votes": null
    },
    {
      "id": "1021094",
      "postDate": "09/21/2020 16:03:45",
      "content": "<p>Thank you!</p>",
      "rawMarkdown": "Thank you!",
      "votes": null
    },
    {
      "id": "1021112",
      "postDate": "09/21/2020 16:25:57",
      "content": "<p>Thank you! </p>\n<p>It looks like ECEF is a 3d coordinate system and not defined only on the surface of earth. However, in this dataset, we will only see surface points. </p>\n<p>I think you are right. z will specify latitude. x and y are sufficient to specify longitude.</p>",
      "rawMarkdown": "Thank you! \n\nIt looks like ECEF is a 3d coordinate system and not defined only on the surface of earth. However, in this dataset, we will only see surface points. \n\nI think you are right. z will specify latitude. x and y are sufficient to specify longitude.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1019059,
      "author_name": "n3n77i",
      "author_url": "",
      "post_date": "09/20/2020 07:03:25",
      "content": "<p>I have gathered that ecef?/enu is used for the protobuf data (lanes, traffic lights), world is the zarr format (ego, agents) and image space is kind of user defined </p>\n<p>There are a couple of hard-coded transform matrices here: <a href=\"https://github.com/lyft/l5kit/blob/master/l5kit/l5kit/rasterization/rasterizer_builder.py\" target=\"_blank\">https://github.com/lyft/l5kit/blob/master/l5kit/l5kit/rasterization/rasterizer_builder.py</a>  see world_to_ecef and ecef_to_aerial</p>\n<p>A bit more info on ecef<br>\n<a href=\"https://www.oc.nps.edu/oc2902w/coord/coord.pdf\" target=\"_blank\">https://www.oc.nps.edu/oc2902w/coord/coord.pdf</a></p>\n<p>Isn't it a 3d surface coordinate system? (both x and y are longitude/ z is lat?)</p>",
      "votes": null,
      "replies": [
        {
          "id": 1021112,
          "author_name": "cmiles715",
          "author_url": "",
          "post_date": "09/21/2020 16:25:57",
          "content": "<p>Thank you! </p>\n<p>It looks like ECEF is a 3d coordinate system and not defined only on the surface of earth. However, in this dataset, we will only see surface points. </p>\n<p>I think you are right. z will specify latitude. x and y are sufficient to specify longitude.</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1020627,
      "author_name": "lucabergamini",
      "author_url": "",
      "post_date": "09/21/2020 10:06:07",
      "content": "<p>This is probably the best resource we have: <a href=\"https://github.com/lyft/l5kit/blob/master/coords_systems.md\" target=\"_blank\">https://github.com/lyft/l5kit/blob/master/coords_systems.md</a></p>",
      "votes": null,
      "replies": [
        {
          "id": 1021094,
          "author_name": "cmiles715",
          "author_url": "",
          "post_date": "09/21/2020 16:03:45",
          "content": "<p>Thank you!</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1018922": "I haven't been able to find definitions for the different coordinate systems used in the l5kit library. What are the ecef, world, image_space, and aerial coordinate systems? what are the origins of each? units of each? and directions of the axes of each?  Also, what is the relationships between these coordinate systems?",
    "1019059": "I have gathered that ecef?/enu is used for the protobuf data (lanes, traffic lights), world is the zarr format (ego, agents) and image space is kind of user defined \n\nThere are a couple of hard-coded transform matrices here: https://github.com/lyft/l5kit/blob/master/l5kit/l5kit/rasterization/rasterizer_builder.py  see world_to_ecef and ecef_to_aerial\n\nA bit more info on ecef\nhttps://www.oc.nps.edu/oc2902w/coord/coord.pdf\n\nIsn't it a 3d surface coordinate system? (both x and y are longitude/ z is lat?)",
    "1020627": "This is probably the best resource we have: https://github.com/lyft/l5kit/blob/master/coords_systems.md",
    "1021094": "Thank you!",
    "1021112": "Thank you! \n\nIt looks like ECEF is a 3d coordinate system and not defined only on the surface of earth. However, in this dataset, we will only see surface points. \n\nI think you are right. z will specify latitude. x and y are sufficient to specify longitude."
  },
  "source": "meta"
}