Abstract: Tools can generate simulated driving tests that allow a same driving scenario to be tested at different geographical locations and with different arrangements of simulated road user(s) and/or simulated road objects. Numerous parameters used by these tools for creating the simulated driving tests can impact the usefulness of the simulated driving tests. If the parameters are not set appropriately, the simulated driving tests may not result in critical interactions that would enable accurate assessment of vehicle software. To alleviate some of these concerns, a parameter tuning system can find optimal value ranges of parameters based on criticality scores of generated simulated driving tests, so that better simulated driving tests can be obtained.

Abstract: The disclosed technology provides solutions for improving simulation scenario generation and in particular, provides methods for improving the generation of simulation scenarios based on autonomous vehicle (AV) driving data. A method of the disclosed technology can include steps for receiving driving data, which includes sensor data from an AV that is descriptive of an environment around the AV, identifying a vehicle in the environment based on the driving data, generating a synthetic driving scenario emulating the environment based on the driving data, and identifying a trajectory of the vehicle in the synthetic driving scenario. The method can further include steps for replacing the vehicle with the AV in the synthetic driving scenario by simulating navigation of the AV in the synthetic driving scenario along the identified trajectory. Systems and machine-readable media are also provided.

Abstract: A track based moving object association is described for distributed sensing applications, such as for identifying multiple objects within a vehicle transportation network for use by a vehicle navigating the network. Position information for the multiple objects within a portion of the vehicle transportation network is obtained from multiple sensors within the portion of the vehicle transportation network. Using the position information of respective sensors of the multiple sensors, a respective track for objects of the multiple objects is determined. Similarity measures are determined for multiple tracks, including at least a first track determined using the position information of a first sensor of the multiple sensors and a second track determined using the position information of a second sensor of the multiple sensors. Based on the similarity measures of the tracks, a tracked object track for a tracked object of the multiple objects is determined.

Abstract: Tools can generate simulated driving tests that allow a same driving scenario to be tested at different geographical locations and with different arrangements of simulated road user(s) and/or simulated road objects. Numerous parameters used by these tools for creating the simulated driving tests can impact the usefulness of the simulated driving tests. If the parameters are not set appropriately, the simulated driving tests may not result in critical interactions that would enable accurate assessment of vehicle software. To alleviate some of these concerns, a parameter tuning system can find optimal value ranges of parameters based on criticality scores of generated simulated driving tests, so that better simulated driving tests can be obtained.

Abstract: Aspects of the disclosed technology provide solutions for identifying autonomous vehicle (AV) tests that provide a desired level of test coverage for testing or validating the AV software stack. A process of the disclosed technology can include steps for extracting a first set of features associated with a first set of test programs, tagging each respective test program with metadata tags, and identifying a second set of features associated with an updated set of AV program code. In some aspects, the process may further include steps for determining if the one or more tags match one or more features of the second set of features associated with the updated AV program code, and executing the respective test programs based on the one or more tags. Systems and machine-readable media are also provided.

Abstract: The disclosed technology provides solutions for improving autonomous vehicle (AV) testing and in particular, for improving AV adversarial testing using offline reinforcement learning. In some aspects, the disclosed technology includes a process for improving AV adversarial testing, including steps for receiving driving data from a database, the driving data representing a plurality of driving scenarios encountered by AVs and training an offline reinforcement learning agent with the driving data. Further, the process includes steps for receiving a driving scenario from the database, calculating a first safety score for the driving scenario, providing the driving scenario to an offline reinforcement learning model, and generating a synthetic scenario for simulating navigation of an AV, the synthetic scenario having a second safety score that is lower than the first safety score.

Abstract: The disclosed technology provides solutions for improving simulation scenario generation and in particular, provides methods for improving the generation of simulation scenarios based on autonomous vehicle (AV) driving data. A method of the disclosed technology can include steps for receiving driving data, which includes sensor data from an AV that is descriptive of an environment around the AV, identifying a vehicle in the environment based on the driving data, generating a synthetic driving scenario emulating the environment based on the driving data, and identifying a trajectory of the vehicle in the synthetic driving scenario. The method can further include steps for replacing the vehicle with the AV in the synthetic driving scenario by simulating navigation of the AV in the synthetic driving scenario along the identified trajectory. Systems and machine-readable media are also provided.