Simulations with modified agents for testing autonomous vehicle software
The disclosure relates to testing software for operating an autonomous vehicle. In one instance, a simulation may be run using log data collected by a vehicle operating in an autonomous driving mode. The simulation may be run using the software to control a simulated vehicle and by modifying a characteristic of an agent identified in the log data. During the running of the simulation, that a first type of interaction between the first simulated vehicle and the modified agent will occur may be determined. In response to determining that the particular type of interaction will occur, the modified agent may be replaced by a interactive agent that simulates a road user corresponding to the modified agent that is capable of responding to actions performed by simulated vehicles. That the particular type of interaction between the simulated vehicle and the interactive agent has occurred in the simulation may be determined.
1. A method of testing software for operating a vehicle in an autonomous driving mode, the method comprising:
retrieving, by one or more processors, log data collected by the vehicle while operating in the autonomous driving mode;
generating, by the one or more processors, a modified version of the log data by modifying an agent identified by the log data;
running, by the one or more processors, a first simulation based on the modified version of the log data using the software to control a simulated vehicle;
during the running of the first simulation, determining, by the one or more processors, whether an interaction between the simulated vehicle and the modified agent has occurred; and
in response to determining that the interaction between the simulated vehicle and the modified agent has occurred, running, by the one or more processors, a second simulation after replacing the modified agent with an interactive agent that simulates a road user corresponding to the modified agent that is capable of responding to actions performed by the simulated vehicle.
2. The method of claim 1 , further comprising:
during the running of the second simulation, determining, by the one or more processors, whether the interaction between the simulated vehicle and the interactive agent has occurred.
3. The method of claim 1 , wherein the interaction includes swerving by the simulated vehicle.
4. The method of claim 1 , wherein the interaction includes hard braking or a particular rate of deceleration by the simulated vehicle.
5. The method of claim 1 , wherein the interaction includes risky behaviors by the simulated vehicle that cause the interactive agent to swerve.
6. The method of claim 1 , wherein the interaction includes risky behaviors by the interactive agent that cause the simulated vehicle to brake hard.
7. The method of claim 1 , wherein the interaction includes a collision or a near collision between the simulated vehicle and the interactive agent.
8. The method of claim 7 , further comprising, in response to determining that the collision or the near collision between the simulated vehicle and the interactive agent has occurred, flagging the second simulation for further review.
9. The method of claim 1 , further comprising identifying the log data by analyzing various log data to identify a situation in which the vehicle operating in the autonomous driving mode is exhibiting a first type of behavior and the agent is exhibiting a second type of behavior.
10. The method of claim 9 , wherein the first type of behavior is changing lanes or making an unprotected turn.
11. The method of claim 10 , wherein the second type of behavior is passing the vehicle operating in the autonomous driving mode.
12. The method of claim 10 , wherein the second type of behavior is exceeding a speed limit for the agent while approaching the vehicle operating in the autonomous driving mode from an oncoming lane.
13. The method of claim 1 , wherein a characteristic of the agent includes a dimension or a shape of the agent.
14. A system for testing software for operating a vehicle in an autonomous driving mode, the system comprising:
one or more processors, wherein the one or more processors are configured to:
retrieve log data collected by the vehicle while operating in the autonomous driving mode;
generate a modified version of the log data by modifying an agent identified by the log data;
run a first simulation based on the modified version of the log data using the software to control a simulated vehicle;
during the running of the first simulation, determine whether an interaction between the simulated vehicle and the modified agent has occurred; and
in response to determining that the interaction between the simulated vehicle and the modified agent has occurred, run a second simulation after replacing the modified agent with an interactive agent that simulates a road user corresponding to the modified agent that is capable of responding to actions performed by the simulated vehicle.
15. The system of claim 14 , wherein the one or more processors are further configured to, during the running of the second simulation, determine whether the interaction between the simulated vehicle and the interactive agent has occurred.
16. The system of claim 14 , wherein the interaction includes swerving by the simulated vehicle.
17. The system of claim 14 , wherein the interaction includes hard braking or a particular rate of deceleration by the simulated vehicle.
18. The system of claim 14 , wherein the interaction includes risky behaviors by the simulated vehicle that cause the interactive agent to swerve.
19. The system of claim 14 , wherein the interaction includes risky behaviors by the interactive agent that cause the simulated vehicle to brake hard.
20. The system of claim 14 , wherein the interaction includes a collision or a near collision between the simulated vehicle and the interactive agent.