Nvidia – assessing pedestrian crossing risk

Intéressant le mode de calcul, mais soit doit necessité énormément de GPU. Sans compter la vision partielle.

Using cameras, LIDAR data and machine learning, vehicles will assess pedestrian risk data. For instance, if there’s an elderly person crossing the road, Nvidia might pick up that the pedestrian has a hearing aid and a walking stick. Nvidia would then calculate the trajectory of the pedestrian and assess if they’re in danger. If a certain risk threshold is met, the vehicle would then emit a beamformed audio signal at the direction of the elderly person, to ensure that the elderly person is aware of the risk.

According to the filing, Nvidia’s system would take into account a number of inputs for assessing risk, such as:

pedestrian’s head pose

  • body pose
  • clothing
  • any articles on a person (e.g. wearing headphones?)
  • gestures
  • weather conditions
  • time of day

via Patent Drop : lire l’article source

Laisser un commentaire

Votre adresse e-mail ne sera pas publiée.

Ce site utilise Akismet pour réduire les indésirables. En savoir plus sur comment les données de vos commentaires sont utilisées.