{
  "id": 179045,
  "title": "One Thousand and One Hours: Self-driving Motion Prediction Dataset",
  "url": "/competitions/lyft-motion-prediction-autonomous-vehicles/discussion/179045",
  "author_name": "Ji Wong Park",
  "post_date": "2020-09-01T07:49:38.959000",
  "votes": 3,
  "comment_count": 0,
  "views": 0,
  "content": "<p><a href=\"https://arxiv.org/pdf/2006.14480.pdf\" target=\"_blank\">https://arxiv.org/pdf/2006.14480.pdf</a></p>\n<blockquote>\n  <p>Abstract : We  present  the  largest  self-driving  dataset  for  motionprediction to date, with over 1,000 hours of data. This wascollected by a fleet of 20 autonomous vehicles along a fixedroute in Palo Alto, California over a four-month period.  Itconsists of 170,000 scenes, where each scene is 25 secondslong and captures the perception output of the self-drivingsystem, which encodes the precise positions and motions ofnearby  vehicles,  cyclists,  and  pedestrians  over  time.   Ontop  of  this,  the  dataset  contains  a  high-definition  seman-tic map with 15,242 labelled elements and a high-definitionaerial view over the area.  Together with the provided soft-ware  kit,  this  collection  forms  the  largest,  most  completeand  detailed  dataset  to  date  for  the  development  of  self-driving,  machine  learning  tasks  such  as  motion  forecast-ing, planning and simulation.  </p>\n</blockquote>",
  "messages": [
    {
      "id": 993885,
      "postDate": "2020-09-01T07:49:38.960Z",
      "content": "<p><a href=\"https://arxiv.org/pdf/2006.14480.pdf\" target=\"_blank\">https://arxiv.org/pdf/2006.14480.pdf</a></p>\n<blockquote>\n  <p>Abstract : We  present  the  largest  self-driving  dataset  for  motionprediction to date, with over 1,000 hours of data. This wascollected by a fleet of 20 autonomous vehicles along a fixedroute in Palo Alto, California over a four-month period.  Itconsists of 170,000 scenes, where each scene is 25 secondslong and captures the perception output of the self-drivingsystem, which encodes the precise positions and motions ofnearby  vehicles,  cyclists,  and  pedestrians  over  time.   Ontop  of  this,  the  dataset  contains  a  high-definition  seman-tic map with 15,242 labelled elements and a high-definitionaerial view over the area.  Together with the provided soft-ware  kit,  this  collection  forms  the  largest,  most  completeand  detailed  dataset  to  date  for  the  development  of  self-driving,  machine  learning  tasks  such  as  motion  forecast-ing, planning and simulation.  </p>\n</blockquote>",
      "rawMarkdown": "https://arxiv.org/pdf/2006.14480.pdf\n\n> Abstract : We  present  the  largest  self-driving  dataset  for  motionprediction to date, with over 1,000 hours of data. This wascollected by a fleet of 20 autonomous vehicles along a fixedroute in Palo Alto, California over a four-month period.  Itconsists of 170,000 scenes, where each scene is 25 secondslong and captures the perception output of the self-drivingsystem, which encodes the precise positions and motions ofnearby  vehicles,  cyclists,  and  pedestrians  over  time.   Ontop  of  this,  the  dataset  contains  a  high-definition  seman-tic map with 15,242 labelled elements and a high-definitionaerial view over the area.  Together with the provided soft-ware  kit,  this  collection  forms  the  largest,  most  completeand  detailed  dataset  to  date  for  the  development  of  self-driving,  machine  learning  tasks  such  as  motion  forecast-ing, planning and simulation.  ",
      "votes": 3
    }
  ],
  "comments": [],
  "raw_markdown_by_id": {
    "993885": "https://arxiv.org/pdf/2006.14480.pdf\n\n> Abstract : We  present  the  largest  self-driving  dataset  for  motionprediction to date, with over 1,000 hours of data. This wascollected by a fleet of 20 autonomous vehicles along a fixedroute in Palo Alto, California over a four-month period.  Itconsists of 170,000 scenes, where each scene is 25 secondslong and captures the perception output of the self-drivingsystem, which encodes the precise positions and motions ofnearby  vehicles,  cyclists,  and  pedestrians  over  time.   Ontop  of  this,  the  dataset  contains  a  high-definition  seman-tic map with 15,242 labelled elements and a high-definitionaerial view over the area.  Together with the provided soft-ware  kit,  this  collection  forms  the  largest,  most  completeand  detailed  dataset  to  date  for  the  development  of  self-driving,  machine  learning  tasks  such  as  motion  forecast-ing, planning and simulation.  "
  }
}