3 ms·

There was a big note in the GTC keynote this year about Nvidia using virtual environments to debug their autopilot algorithms. Think "car" being driven in a hig

by ethbro 9y ago

There was a big note in the GTC keynote this year about Nvidia using virtual environments to debug their autopilot algorithms. Think "car" being driven in a high fidelity video game. Advantage being that if you can provide representative, simulated input then you can increase the training miles by orders of magnitude in the same amount of time, limited by computation rather than physical mileage.

firebones 9y ago
This is true, to the extent that the selection of input parameters to simulate provides coverage over the domain of all possible input parameters. At some point you're back to testing the failure of imagination of the simulation/test creators.
throwaway2048 9y ago
The problem is the simulation by definition can only simulate things that are accounted for. Any number of completely arbitrary out of the blue things can happen in real life.
- nitrogen 9y ago
  Simulations can make use of random number generators and could in theory, following the lead of a project like afl, adaptively find algorithmic weak spots.
  - taneq 9y ago
    Oooh, adversarial scenario generation! I think there's something to that.
  - sah2ed 9y ago
    Simulations are only useful up to a point. Trying to use them to model real world scenarios would be useless in practice, due to the Ludic fallacy [0]. Real life is too complex to be modeled in any simulation. [0] https://en.wikipedia.org/wiki/Ludic_fallacy https://en.wikipedia.org/wiki/Ludic_fallacy
    - nitrogen 9y ago
      I don't think the connection to the referenced fallacy is nearly strong enough to serve as a QED on its own. It's also seemingly promoted only by one person. As for modeling real life with simulations, the data exists for every type of accident a human has encountered. If you incorporate such into a simulation, plus randomly vary every free parameter, then your simulation will cover more scenarios than any human driver can possibly experience. Thus, simulations should be able to help an autonomous vehicle outperform humans by a large margin, which is the only goal that matters.

by ethbro 9y ago

firebones 9y ago

This is true, to the extent that the selection of input parameters to simulate provides coverage over the domain of all possible input parameters. At some point you're back to testing the failure of imagination of the simulation/test creators.

throwaway2048 9y ago

The problem is the simulation by definition can only simulate things that are accounted for. Any number of completely arbitrary out of the blue things can happen in real life.

nitrogen 9y ago
Simulations can make use of random number generators and could in theory, following the lead of a project like afl, adaptively find algorithmic weak spots.
- taneq 9y ago
  Oooh, adversarial scenario generation! I think there's something to that.
- sah2ed 9y ago
  Simulations are only useful up to a point. Trying to use them to model real world scenarios would be useless in practice, due to the Ludic fallacy [0]. Real life is too complex to be modeled in any simulation. [0] https://en.wikipedia.org/wiki/Ludic_fallacy https://en.wikipedia.org/wiki/Ludic_fallacy
  - nitrogen 9y ago
    I don't think the connection to the referenced fallacy is nearly strong enough to serve as a QED on its own. It's also seemingly promoted only by one person. As for modeling real life with simulations, the data exists for every type of accident a human has encountered. If you incorporate such into a simulation, plus randomly vary every free parameter, then your simulation will cover more scenarios than any human driver can possibly experience. Thus, simulations should be able to help an autonomous vehicle outperform humans by a large margin, which is the only goal that matters.