Machine Learning for Autonomous Vehicle Perception at Cruise




[ad_1]

Sean Harris discusses one of the key challenges in AV development–prediction. AV prediction focuses on estimating what other agents (e.g. drivers, pedestrians, cyclists) are going to do next. This problem is critical for ensuring our AVs drive safely, but it isn’t a challenge that’s common in many other robotics or ML domains. This is one reason why AVs push machine learning to the edge.

Source


[ad_2]

Comment List

  • Cruise
    December 20, 2020

    sounds like a terrible way to label the dataset. You should use physics based labels as much as possible. Just predict the trajectories for example.

  • Cruise
    December 20, 2020

    Nice crisp presentation Sean👍, I have a question? As you were mentioning about identifying the edge case , labelling that case and feeding to the model again , so I wanted to ask how often do you retrain the same model? We might come across a edge case very often as the real world is stochastic.

  • Cruise
    December 20, 2020

    Never going to happen. The first fatality will be the last. What a scum bag company. They have an office full of employees in their SF office. Completely illegal. non essential company.

  • Cruise
    December 20, 2020

    hi I got a question about self driving car video, is there any inside view? only we can see the view of the lense

  • Cruise
    December 20, 2020
  • Cruise
    December 20, 2020

    Nice presentation Sean! If I could go back in time I would've put more effort in the AI tutorials you lectured to my class at UNSW. Great job!

Write a comment