Injecting noise into robot simulation
Implementations are provided for increasing realism of robot simulation by injecting noise into various aspects of the robot simulation. In various implementations, a three-dimensional (3D) environment may be simulated and may include a simulated robot controlled by an external robot controller. Joint command(s) issued by the robot controller and/or simulated sensor data passed to the robot controller may be intercepted. Noise may be injected into the joint command(s) to generate noisy commands. Additionally or alternatively, noise may be injected into the simulated sensor data to generate noisy sensor data. Joint(s) of the simulated robot may be operated in the simulated 3D environment based on the one or more noisy commands. Additionally or alternatively, the noisy sensor data may be provided to the robot controller to cause the robot controller to generate joint commands to control the simulated robot in the simulated 3D environment.
1. A method implemented using one or more processors, comprising:
simulating an environment, wherein the simulated environment includes a simulated robot controlled by an external robot controller;
intercepting a temporal sequence of robot commands issued by the robot controller to control the simulated robot in the simulated environment;
generating a temporal sequence of progressing noise values using a time series machine learning model, wherein the time series machine learning model is trained based on empirical data about effects of environmental conditions on robot joint operation;
injecting the temporal sequence of progressing noise values into corresponding robot commands of the temporal sequence of robot commands to generate noisy robot commands, wherein the injecting includes altering, skewing, biasing, or replacing existing robot command values with noisy robot command values; and
operating the simulated robot in the simulated environment based on the noisy robot commands.
2. The method of claim 1 , wherein injecting the temporal sequence of progressing noise values includes:
operating a state machine to generate the temporal sequence of progressing noise values, wherein one or more progressing noise values of the temporal sequence of progressing noise values are generated based at least in part on one or more previous progressing noise values of the temporal sequence of progressing noise values.
3. The method of claim 1 , further comprising generating the temporal sequence of progressing noise values using a time series machine learning model.
4. The method of claim 3 , wherein the generating comprises iteratively applying time series data indicative of environmental conditions as inputs across the time series machine learning model.
5. A method implemented using one or more processors, comprising:
simulating an environment, wherein the simulated environment includes a simulated robot controlled by an external robot controller;
intercepting a temporal sequence of robot commands issued by the robot controller to control the simulated robot in the simulated environment;
generating a temporal sequence of progressing noise values using a time series machine learning model, wherein the generating comprises iteratively applying time series data indicative of wear and tear metrics associated with a real-world robot;
injecting the temporal sequence of progressing noise values into corresponding robot commands of the temporal sequence of robot commands to generate noisy robot commands, wherein the injecting includes altering, skewing, biasing, or replacing existing robot command values with noisy robot command values; and
operating the simulated robot in the simulated environment based on the noisy robot commands.
6. A method implemented using one or more processors, comprising:
simulating an environment, wherein the simulated environment includes a simulated robot controlled by an external robot controller;
intercepting a temporal sequence of robot commands issued by the robot controller to control the simulated robot in the simulated environment;
injecting a temporal sequence of progressing noise values into corresponding robot commands of the temporal sequence of robot commands to generate noisy robot commands, wherein the injecting includes altering, skewing, biasing, or replacing existing robot command values with noisy robot command values, and wherein the temporal sequence of progressing noise values simulate a bias of gravity over time on oscillations of one or more joints of the simulated robot; and
operating the simulated robot in the simulated environment based on the noisy robot commands.
7. The method of claim 1 , wherein the temporal sequence of progressing noise values are calculated based on one or more operating tolerances of a robot controlled by the robot controller.
8. The method of claim 7 , wherein one or more of the operating tolerances include a motor operating tolerance.
9. A system comprising one or more processors and memory storing instructions that, in response to execution by the one or more processors, cause the one or more processors to:
simulate an environment, wherein the simulated environment includes a simulated robot controlled by an external robot controller;
intercept a temporal sequence of robot commands issued by the robot controller to control the simulated robot in the simulated environment;
generate a temporal sequence of progressing noise values using a time series machine learning model, wherein the time series machine learning model is trained based on empirical data about effects of environmental conditions on robot joint operation;
inject the temporal sequence of progressing noise values into corresponding robot commands of the temporal sequence of robot commands to generate noisy robot commands, wherein the injecting includes altering, skewing, biasing, or replacing existing robot command values with noisy robot command values; and
operate the simulated robot in the simulated environment based on the noisy robot commands.
10. The system of claim 9 , wherein the instructions to inject the temporal sequence of progressing noise values include instructions to:
operate a state machine to generate the temporal sequence of progressing noise values, wherein one or more progressing noise values of the temporal sequence of progressing noise values are generated based at least in part on one or more previous progressing noise values of the temporal sequence of progressing noise values.
11. The system of claim 9 , further comprising instructions to generate the temporal sequence of progressing noise values using a time series machine learning model.
12. The system of claim 11 , wherein the instructions to generate include instructions to iteratively apply time series data indicative of environmental conditions as inputs across the time series machine learning model.
13. A system comprising one or more processors and memory storing instructions that, in response to execution by the one or more processors, cause the one or more processors to:
simulate an environment, wherein the simulated environment includes a simulated robot controlled by an external robot controller;
intercept a temporal sequence of robot commands issued by the robot controller to control the simulated robot in the simulated environment;
generate a temporal sequence of progressing noise values using a time series machine learning model, wherein the instructions to generate include instructions to iteratively apply time series data indicative of wear and tear metrics associated with a real-world robot;
inject the temporal sequence of progressing noise values into corresponding robot commands of the temporal sequence of robot commands to generate noisy robot commands, wherein the injecting includes altering, skewing, biasing, or replacing existing robot command values with noisy robot command values; and
operate the simulated robot in the simulated environment based on the noisy robot commands.
14. A system comprising one or more processors and memory storing instructions that, in response to execution by the one or more processors, cause the one or more processors to:
simulate an environment, wherein the simulated environment includes a simulated robot controlled by an external robot controller;
intercept a temporal sequence of robot commands issued by the robot controller to control the simulated robot in the simulated environment;
inject a temporal sequence of progressing noise values into corresponding robot commands of the temporal sequence of robot commands to generate noisy robot commands, wherein the injecting includes altering, skewing, biasing, or replacing existing robot command values with noisy robot command values, and wherein the temporal sequence of progressing noise values simulate a bias of gravity over time on oscillations of one or more joints of the simulated robot; and
operate the simulated robot in the simulated environment based on the noisy robot commands.
15. The system of claim 9 , wherein the temporal sequence of progressing noise values are calculated based on one or more operating tolerances of a robot controlled by the robot controller.
16. The system of claim 15 , wherein one or more of the operating tolerances include a motor operating tolerance.