Detecting and responding to sounds for autonomous vehicles
The technology relates to detecting and responding to sounds for a vehicle having an autonomous driving mode. In one example, an audible signal corresponding to a sound received at one or more microphones of the vehicle may be received. Sensor data generated by a perception system of the vehicle identifying objects in an environment of the vehicle may be received. A type of sound may be determined by inputting the audible signal into a classifier. A set of additional signals may be determined based on the determined type of sound. The sensor data may be processed in order to identify one or more additional signals of the identified set of additional signals. The vehicle may be controlled in the autonomous driving mode in order to respond to the sound based on the one or more additional signals and the type of sound.
1. A method of detecting and responding to sounds for a vehicle having an autonomous driving mode, the method comprising:
receiving, by one or more processors, an audible signal corresponding to a sound received at one or more microphones of the vehicle;
receiving, by the one or more processors, sensor data generated by a perception system of the vehicle identifying objects in an environment of the vehicle, the perception system including one or more sensors;
inputting, by the one or more processors, the audible signal into a classifier;
determining, by the classifier, a type of sound associated with the inputted audible signal;
outputting, by the classifier, a likelihood value for the determined type of sound;
controlling, by the one or more processors, the vehicle in the autonomous driving mode in order to initially respond to the sound based on the determined type of sound;
determining, by the one or more processors, whether the likelihood value meets a first threshold;
when the likelihood value is determined to meet the first threshold:
identifying, by the one or more processors, a set of additional signals based on the determined type of sound;
processing, by the one or more processors, the sensor data in order to identify one or more additional signals of the set of additional signals;
adjusting, by the one or more processors, the likelihood value based on the identified one or more additional signals; and
determining, by the one or more processors, whether the adjusted likelihood value meets a second threshold; and
when the likelihood value is determined to meet the second threshold:
controlling, by the one or more processors, the vehicle in the autonomous driving mode in order to continue to respond to the sound based on the identified one or more additional signals and the determined type of sound.
2. The method of claim 1 , wherein the likelihood value is increased based on the identified one or more additional signals.
3. The method of claim 2 , wherein the second threshold is greater than the first threshold.
4. The method of claim 1 , further comprising, determining that the one or more additional signals are a predetermined combination of additional signals, and wherein controlling the vehicle is further based on the determination that the one or more additional signals are the predetermined combination of additional signals.
5. The method of claim 1 , further comprising training the classifier using examples of sounds relevant to driving decisions.
6. The method of claim 5 , wherein sounds relevant to driving include sounds which would likely cause a change in behavior of the vehicle.
7. The method of claim 1 , wherein the determined type of sound is a train whistle and the one or more additional signals includes one or more of a flashing light, a gate, a train, a train station identified in pre-stored map information of the vehicle within a predetermined distance of a current location of the vehicle, or a railroad crossing identified in pre-stored map information of the vehicle within a predetermined distance of a current location of the vehicle.
8. The method of claim 1 , wherein the determined type of sound is a reverse beeping sound, and the one or more additional signals includes one or more of a vehicle of a given size or flashing lights.
9. The method of claim 1 , wherein the determined type of sound is a crosswalk chirp, and the one or more additional signals includes one or more of a crosswalk, a walk sign, or flashing lights.
10. The method of claim 1 , further comprising, prior to identifying the one or more additional signals, controlling the vehicle in the autonomous driving mode in order to perform an initial response based the determined type of sound.
11. The method of claim 1 , wherein the classifier is stored by a memory.
12. A system for detecting and responding to sounds for a vehicle having an autonomous driving mode, the system comprising:
one or more processors configured to:
receive an audible signal corresponding to a sound received at one or more microphones of the vehicle;
receive sensor data generated by a perception system of the vehicle identifying objects in an environment of the vehicle, the perception system including one or more sensors;
input the audible signal into a classifier;
determine a type of sound associated with the inputted audible signal;
output from the classifier a likelihood value for the determined type of sound;
control the vehicle in the autonomous driving mode in order to initially respond to the sound based on the determined type of sound; and
determine whether the likelihood value meets a first threshold;
when the likelihood value is determined to meet the first threshold:
identify a set of additional signals based on the determined type of sound;
process the sensor data in order to identify one or more additional signals of the set of additional signals;
adjust the likelihood value based on the identified one or more additional signals; and
determine whether the adjusted likelihood value meets a second threshold; and
when the likelihood value is determined to meet the second threshold:
control the vehicle in the autonomous driving mode in order to continue to respond to the sound based on the identified one or more additional signals and the determined type of sound.
13. The system of claim 12 , wherein the likelihood value is increased based on the identified one or more additional signals.
14. The system of claim 13 , wherein the second threshold is greater than the first threshold.
15. The system of claim 12 , wherein the one or more processors are further configured to determine that the one or more additional signals are a predetermined combination of additional signals, and wherein controlling the vehicle is further based on the determination that the one or more additional signals are the predetermined combination of additional signals.
16. The system of claim 12 , wherein the determined type of sound is a railroad warning bell and the one or more additional signals includes one or more of a flashing light, a gate, a train, a train station identified in pre-stored map information of the vehicle within a predetermined distance of a current location of the vehicle, or a railroad crossing identified in pre-stored map information of the vehicle within a predetermined distance of a current location of the vehicle.
17. The system of claim 12 , wherein the determined type of sound is a reverse beeping sound, and the one or more additional signals includes one or more of a vehicle of a given size or flashing lights.
18. The system of claim 12 , wherein the determined type of sound is a crosswalk chirp, and the one or more additional signals includes one or more of a crosswalk, a walk sign, or flashing lights.
19. The system of claim 12 , further comprising the vehicle, the perception system, and the one or more microphones.
20. The system of claim 12 , wherein the classifier is stored by a memory.