Autonomous vehicle model training and validation using low-discrepancy sequences
Autonomous vehicle model training and validation using low-discrepancy sequences may include: generating a low-discrepancy sequence in a multidimensional space comprising a plurality of multidimensional points; mapping each sample of a plurality of samples of a data corpus to a corresponding entry in the low-discrepancy sequence, wherein each sample of the plurality of samples comprises one or more environmental descriptors for an environment relative to a vehicle and one or more state descriptors describing a state of the vehicle; selecting, from the data corpus, a training data set by selecting, for each multidimensional point of the low-discrepancy sequence having one or more mapped samples, a mapped sample for inclusion in the training data set; and training one or more models used to generate autonomous driving decisions of an autonomous vehicle based on the selected training data set.
1. A method comprising:
receiving a request to generate a synthetic data set comprising a number of samples, wherein each sample of the synthetic data set is encoded as record comprising a plurality of field-value pairs, wherein the plurality of field-value pairs comprise one or more environmental descriptors for an environment relative to a vehicle and one or more state descriptors describing a state of the vehicle;
generating, based on the number of samples, a low-discrepancy sequence in a multidimensional space comprising a plurality of multidimensional points having a number of dimensions equal to a number of continuous fields in the synthetic data set;
generating the synthetic data set based on the low-discrepancy sequence by generating, for each entry in the low-discrepancy sequence, one or more samples of the synthetic data set; and
validating one or more autonomous driving models using a plurality of simulations each initialized using a corresponding sample in the synthetic data set.
2. The method of claim 1 , wherein each sample of the synthetic data set comprises a plurality of continuous fields each corresponding to a dimension of the multidimensional space.
3. The method of claim 2 , wherein generating the synthetic data set comprises converting, for one or more values in each of the plurality of multidimensional points, the one or more values to one or more corresponding continuous values in the synthetic data set.
4. The method of claim 2 , wherein each sample in the synthetic data set further comprises one or more categorical fields.
5. The method of claim 4 , wherein generating the synthetic data set comprises generating, for each combination of categorical field values of the one or more categorical fields, a subset of the synthetic data set based on the low-discrepancy sequence.
6. The method of claim 4 , wherein the one or more categorical fields comprises one or more of: a lane position of an ego vehicle or a number of other vehicles relative to the ego vehicle.
7. The apparatus of claim 4 , wherein generating the synthetic data set comprises generating, for each combination of categorical field values of the one or more categorical fields, a subset of the synthetic data set based on the low-discrepancy sequence.
8. The apparatus of claim 4 , wherein the one or more categorical fields comprises one or more of: a lane position of an ego vehicle or a number of other vehicles relative to the ego vehicle.
9. The method of claim 2 , wherein the plurality of continuous fields comprise one or more of: one or more vehicle behavior characteristics of an ego vehicle, a road characteristic, a position of one or more other vehicles relative to the ego vehicle, or one or more behavioral characteristics for the one or more other vehicles.
10. The apparatus of claim 2 , wherein generating the synthetic data set comprises converting, for one or more values in each of the plurality of multidimensional points, the one or more values to one or more corresponding continuous values in the synthetic data set.
11. The apparatus of claim 2 , wherein each sample in the synthetic data set further comprises one or more categorical fields.
12. The apparatus of claim 2 , wherein the plurality of continuous fields comprise one or more of: one or more vehicle behavior characteristics of an ego vehicle, a road characteristic, a position of one or more other vehicles relative to the ego vehicle, or one or more behavioral characteristics for the one or more other vehicles.
13. The method of claim 1 , wherein validating the one or more autonomous driving models comprises comparing one or more results for each of the plurality of simulations to one or more expected results.
14. The method of claim 1 , wherein validating the one or more autonomous driving models comprises comparing one or more results for each of the plurality of simulations to one or more satisfactory conditions.
15. The method of claim 1 , wherein the low-discrepancy sequence comprises a Sobol sequence.
16. The apparatus of claim 1 , wherein each sample of the synthetic data set comprises a plurality of continuous fields each corresponding to a dimension of the multidimensional space.
17. The apparatus of claim 1 , wherein validating the one or more autonomous driving models comprises comparing one or more results for each of the plurality of simulations to one or more expected results.
18. The apparatus of claim 1 , wherein validating the one or more autonomous driving models comprises comparing one or more results for each of the plurality of simulations to one or more satisfactory conditions.
19. The apparatus of claim 1 , wherein the low-discrepancy sequence comprises a Sobol sequence.
20. An apparatus comprising at least one processor and memory storing instructions that, when executed, cause the at least one processor to perform steps comprising:
receiving a request to generate a synthetic data set comprising a number of samples, wherein each sample of the synthetic data set is encoded as record comprising a plurality of field-value pairs, wherein the plurality of field-value pairs comprise one or more environmental descriptors for an environment relative to a vehicle and one or more state descriptors describing a state of the vehicle;
generating, based on the number of samples, a low-discrepancy sequence in a multidimensional space comprising a plurality of multidimensional points having a number of dimensions equal to a number of continuous fields in the synthetic data set;
generating the synthetic data set based on the low-discrepancy sequence by generating, for each entry in the low-discrepancy sequence, one or more samples of the synthetic data set; and
validating one or more autonomous driving models using a plurality of simulations each initialized using a corresponding sample in the synthetic data set.