Systems and methods for generation and utilization of vehicle testing knowledge structures for autonomous vehicle simulation

1. A computer-implemented method, comprising:

obtaining, by a computing system comprising one or more computing devices, a first vehicle testing tuple comprising a plurality of first testing parameters and a second vehicle testing tuple comprising a plurality of second testing parameters;

determining, by the computing system, that each of the plurality of first testing parameters are associated with an evaluated operating condition;

appending, by the computing system at a first time step, the first vehicle testing tuple to a first portion of a plurality of portions of a vehicle testing knowledge structure, wherein each of the plurality of portions comprises one or more vehicle testing tuples, wherein the plurality of portions corresponds to a respective plurality of time steps, wherein the first portion corresponds to the first time step;

determining, by the computing system, that a second testing parameter of the plurality of second testing parameters is associated with an unevaluated operating condition;

evaluating, by the computing system at a second time step after the first time step, the unevaluated operating condition associated with the second testing parameter; and

generating, by the computing system, a second portion of the vehicle testing knowledge structure comprising the second vehicle testing tuple for the vehicle testing knowledge structure, wherein the second portion of the vehicle testing knowledge structure corresponds to the second time step, wherein the second portion of the vehicle testing knowledge structure is subsequent to the first portion of the vehicle testing knowledge structure, and wherein the second portion of the vehicle testing knowledge structure comprises each vehicle testing tuple of the first portion of the vehicle testing knowledge structure.

2. The computer-implemented method of claim 1 , wherein the method further comprises:

obtaining, by the computing system, performance data descriptive of autonomous vehicle performance for a first scenario type;

determining, by the computing system, a subset of portions from the plurality of portions of the vehicle testing knowledge structure, wherein each of the subset of portions comprises one or more vehicle testing tuples associated with the first scenario type; and

performing, by the computing system, simulations using each of the subset of portions to obtain simulation data descriptive of simulated autonomous vehicle performance for the first scenario type.

3. The computer-implemented method of claim 2 , wherein the method further comprises:

evaluating, by the computing system, a difference between the performance data descriptive of the autonomous vehicle performance and the simulation data to determine a simulation accuracy value indicative of a simulation accuracy for the first scenario type by the vehicle testing knowledge structure; and

verifying, by the computing system based at least in part on the simulation accuracy value, the vehicle testing knowledge structure for generation of simulation data equivalent to the data descriptive of the autonomous vehicle performance for the first scenario type.

4. The computer-implemented method of claim 3 , wherein the method further comprises:

reducing, by the computing system, a number of required autonomous vehicle testing miles by a number of miles associated with the simulation data descriptive of the autonomous vehicle performance for the first scenario type.

5. The computer-implemented method of claim 1 , wherein the method further comprises:

obtaining, by the computing system, data descriptive of an unknown scenario type;

determining, by the computing system, that a known scenario type represented by one or more vehicle testing tuples of the vehicle testing knowledge structure is similar to the unknown scenario type; and

simulating, by the computing system, a scenario of the unknown scenario type using the one or more vehicle testing tuples of the vehicle testing knowledge structure associated with the known scenario type to obtain simulation data descriptive of autonomous vehicle performance for the unknown scenario type.

6. The computer-implemented method of claim 5 , wherein determining that the known scenario type represented by the one or more vehicle testing tuples of the vehicle testing knowledge structure is similar to the unknown scenario type comprises:

obtaining, by the computing system, vehicle operating data associated with the unknown scenario type; and

evaluating, by the computing system, a difference between the vehicle operating data and previous performance data associated with simulation of the known scenario type using the one or more vehicle testing tuples.

7. The computer-implemented method of claim 5 , wherein the unknown scenario type comprises a plurality of autonomous vehicle maneuvers, and wherein the known scenario type comprises a subset of the plurality of autonomous vehicle maneuvers.

8. The computer-implemented method of claim 1 , wherein the plurality of first testing parameters describe at least one of:

one or more environmental conditions;

one or more required vehicle maneuvers;

one or more surface conditions;

a location for each of one or more actors;

a pose for each of the one or more actors; or

one or more respective behaviors for each of the one or more actors.

9. The computer-implemented method of claim 1 , wherein:

the unevaluated operation condition comprises an operating condition that has not been evaluated for operation of autonomous vehicles; and

evaluating, by the computing system at the second time step after the first time step, the unevaluated operating condition associated with the second testing parameter comprises:

obtaining, by the computing system, data descriptive of a performance of an autonomous vehicle for the unevaluated operating condition; and

storing, by the computing system based at least in part on the data, the unevaluated operating condition as an evaluated operating condition within the vehicle testing knowledge structure.

10. The computer-implemented method of claim 9 , wherein the data descriptive of a performance of an autonomous vehicle that is operating under the unevaluated operating condition is based at least in part on:

a simulation of the autonomous vehicle for the unevaluated operating condition;

sensor data from the autonomous vehicle for the unevaluated operating condition; or

sensor data from the autonomous vehicle for an evaluated operating condition equivalent to the unevaluated operating condition.

11. A computing system, comprising:

one or more processors;

a vehicle testing knowledge structure comprising a plurality of portions, wherein each of the plurality of portions comprises one or more vehicle testing tuples each comprising a plurality of vehicle testing parameters, wherein the plurality of portions corresponds to a respective plurality of time steps; and

one or more tangible, non-transitory computer readable media storing computer-readable instructions that when executed by the one or more processors cause the one or more processors to perform operations, the operations comprising:

obtaining a first vehicle testing tuple comprising a plurality of first testing parameters and a second vehicle testing tuple comprising a plurality of second testing parameters;

determining that each of the plurality of first testing parameters are associated with an evaluated operating condition;

appending the first vehicle testing tuple to a first portion of the plurality of portions of the vehicle testing knowledge structure at a first time step of the plurality of time steps, wherein the first portion corresponds to the first time step;

determining that a second testing parameter of the plurality of second testing parameters is associated with an unevaluated operating condition;

evaluating the unevaluated operating condition associated with the second testing parameter at a second time step after the first time step; and

generating a second portion of the vehicle testing knowledge structure comprising the second vehicle testing tuple for the vehicle testing knowledge structure, wherein the second portion of the vehicle testing knowledge structure corresponds to the second time step, wherein the second portion of the vehicle testing knowledge structure is subsequent to the first portion of the vehicle testing knowledge structure, and wherein the second portion of the vehicle testing knowledge structure comprises each vehicle testing tuple of the first portion of the vehicle testing knowledge structure.

12. The computing system of claim 11 , wherein the operations further comprise:

obtaining performance data descriptive of autonomous vehicle performance for a first scenario type;

determining a subset of portions from the plurality of portions of the vehicle testing knowledge structure, wherein each of the subset of portions comprises one or more vehicle testing tuples associated with the first scenario type; and

performing simulations using each of the subset of portions to obtain simulation data descriptive of simulated autonomous vehicle performance for the first scenario type.

13. The computing system of claim 12 , wherein the operations further comprise:

evaluating a difference between the performance data descriptive of the autonomous vehicle performance and the simulation data to determine a simulation accuracy value indicative of a simulation accuracy for the first scenario type by the vehicle testing knowledge structure; and

verifying, based at least in part on the simulation accuracy value, the vehicle testing knowledge structure for generation of simulation data equivalent to the data descriptive of the autonomous vehicle performance for the first scenario type.

14. The computing system of claim 13 , wherein the operations further comprise:

reducing a number of required autonomous vehicle testing miles by a number of miles associated with the simulation data descriptive of the autonomous vehicle performance for the first scenario type.

15. The computing system of claim 13 , wherein the operations further comprise:

obtaining data descriptive of an unknown scenario type;

determining that a known scenario type represented by one or more vehicle testing tuples of the vehicle testing knowledge structure is similar to the unknown scenario type; and

simulating a scenario of the unknown scenario type using the one or more vehicle testing tuples of the vehicle testing knowledge structure associated with the known scenario type to obtain simulation data descriptive of autonomous vehicle performance for the unknown scenario type.

16. The computing system of claim 15 , wherein determining that the known scenario type represented by the one or more vehicle testing tuples of the vehicle testing knowledge structure is similar to the unknown scenario type comprises:

obtaining vehicle operating data associated with the unknown scenario type; and

evaluating a difference between the vehicle operating data and previous performance data associated with simulation of the known scenario type using the one or more vehicle testing tuples.

17. The computing system of claim 11 , wherein:

the unevaluated operation condition comprises an operating condition that has not been evaluated for operation of autonomous vehicles; and

evaluating, at the second time step after the first time step, the unevaluated operating condition associated with the second testing parameter comprises:

obtaining data descriptive of a performance of an autonomous vehicle for the unevaluated operating condition; and

storing, based at least in part on the data, the unevaluated operating condition as an evaluated operating condition within the vehicle testing knowledge structure.

18. One or more tangible, non-transitory computer readable media storing computer-readable instructions that when executed by one or more processors cause the one or more processors to perform operations, the operations comprising:

obtaining a first vehicle testing parameter grouping comprising a plurality of first testing parameters and a second vehicle testing parameter grouping comprising a plurality of second testing parameters;

determining that each of the plurality of first testing parameters are associated with an evaluated operating condition;

appending, at a first time step, the first vehicle testing parameter grouping to a first portion of a plurality of portions of a vehicle testing knowledge structure, wherein each of the plurality of portions comprises one or more vehicle testing parameter groupings, wherein the plurality of portions corresponds to a respective plurality of time steps, wherein the first portion corresponds to the first time step;

determining that a second testing parameter of the plurality of second testing parameters is associated with an unevaluated operating condition;

evaluating, at a second time step after the first time step, the unevaluated operating condition associated with the second testing parameter; and

generating a second portion of the vehicle testing knowledge structure comprising the second vehicle testing parameter grouping for the vehicle testing knowledge structure, wherein the second portion of the vehicle testing knowledge structure corresponds to the second time step, wherein the second portion is subsequent to the first portion of the vehicle testing knowledge structure, and wherein the second portion of the vehicle testing knowledge structure comprises each vehicle testing parameter grouping of the first portion of the vehicle testing knowledge structure.

19. The one or more tangible, non-transitory media of claim 18 , wherein the operations further comprise:

obtaining performance data descriptive of autonomous vehicle performance for a first scenario type;

determining a subset of portions from the plurality of portions of the vehicle testing knowledge structure, wherein each of the subset of portions comprises one or more vehicle testing parameter groupings associated with the first scenario type;

performing simulations using each of the subset of portions to obtain simulation data descriptive of simulated autonomous vehicle performance for the first scenario type;

evaluating a difference between the performance data descriptive of the autonomous vehicle performance and the simulation data to determine a simulation accuracy value indicative of a simulation accuracy for the first scenario type by the vehicle testing knowledge structure; and

verifying, based at least in part on the simulation accuracy value, the vehicle testing knowledge structure for generation of simulation data equivalent to the data descriptive of the autonomous vehicle performance for the first scenario type.