| CPC G05D 1/0088 (2013.01) [B60W 60/001 (2020.02); G05B 17/02 (2013.01); G05D 1/0276 (2013.01)] | 20 Claims |
|
1. A method comprising:
receiving an instruction to test a component of an autonomous vehicle by simulating operation of the autonomous vehicle based at least in part on providing simulated data as input to the component of the autonomous vehicle;
determining a simulated scenario based at least in part on selecting a first set of scenario parameters from among multiple scenario parameter sets, wherein determining the first set of scenario parameters comprises:
receiving a first probability distribution associated with different sets of scenario parameters, wherein:
the first probability distribution comprises a probability associated with the first set of scenario parameters,
the probability indicates a likelihood that the autonomous vehicle will encounter a scenario defined by the first set of scenario parameters during operation of the autonomous vehicle, and
the first set of scenario parameters identifies a characteristic of a simulated environment, a simulated object in the environment, and a simulated vehicle associated with the autonomous vehicle;
receiving a classification associated with the first set of scenario parameters indicating that the first set of scenario parameters is associated with an adverse event;
modifying, based at least in part on the classification, the first probability distribution as a modified distribution, wherein the modified distribution increases a likelihood of an adverse event occurring in a simulated scenario and modifies the probability as a modified probability, wherein the modifying is further based at least in part on determining, as a parameterized transformation of the multiple scenario parameter sets, an explicit representation of a parameter space associated with the multiple scenario parameter sets; and
determining, based at least in part on the modified distribution and the modified probability, the first set of scenario parameters from among the different sets of scenario parameters; and
simulating a first scenario based at least in part on the first set of scenario parameters, wherein the simulating comprises simulating operation of the simulated vehicle based at least in part on operating a component of the autonomous vehicle using simulated data generated by simulating the first scenario.
|