Determining traffic control features based on telemetry patterns within digital image representations of vehicle telemetry data
The present disclosure relates to systems, methods, and non-transitory computer readable media for identifying traffic control features based on telemetry patterns within digital image representations of vehicle telemetry information. The disclosed systems can generate a digital image representation based on collected telemetry information to represent the frequency of different speed-location combinations for transportation vehicles passing through a traffic area. The disclosed systems can also apply a convolutional neural network to analyze the digital image representation and generate a predicted classification of a type of traffic control feature that corresponds to the digital image representation of vehicle telemetry information. The disclosed systems further train the convolutional neural network to determine traffic control features based on training data.
1. A method comprising:
generating a training telemetry digital image comprising pixels representing telemetry information for a traffic area;
determining a predicted traffic control feature associated with the traffic area by processing the pixels representing the telemetry information from the training telemetry digital image utilizing a machine learning model; and
training the machine learning model by comparing the predicted traffic control feature to a ground truth traffic control feature associated with the traffic area.
2. The method of claim 1 , wherein training the machine learning model by comparing the predicted traffic control feature to the ground truth traffic control feature comprises determining a measure of loss between the predicted traffic control feature and the ground truth traffic control feature utilizing a loss function.
3. The method of claim 2 , wherein the machine learning model comprises a convolutional neural network and training the machine learning model comprises modifying parameters of the convolutional neural network from the loss function utilizing back propagation.
4. The method of claim 1 , further comprising:
generating an additional training telemetry digital image comprising pixels representing additional telemetry information for an additional traffic area; and
training the machine learning model by comparing, from the pixels of the additional training telemetry digital image, an additional predicted traffic control feature to an additional ground truth traffic control feature associated with the additional traffic area.
5. The method of claim 1 , wherein determining the training telemetry digital image comprises determining speeds and locations of transportation vehicles as the transportation vehicles pass through the traffic area.
6. The method of claim 1 , wherein determining the training telemetry digital image further comprises:
identifying a plurality of speed-location combinations indicating speeds of transportation vehicles at corresponding locations within the traffic area;
determining frequencies with which the plurality of speed-location combinations occur within the traffic area; and
generating pixels of the training telemetry digital image from the frequencies of the plurality of speed-location combinations.
7. The method of claim 6 , wherein generating the pixels comprises:
generating a first pixel of the training telemetry digital image based on a first frequency of a first speed-location combination; and
generating a second pixel of the training telemetry digital image based on a second frequency of a second speed-location combination.
8. The method of claim 1 , further comprising:
generating a telemetry digital image comprising pixels representing additional telemetry information for an additional traffic area; and
generating a traffic control feature associated with the additional traffic area by processing the pixels representing the additional telemetry information from the telemetry digital image utilizing the machine learning model.
9. The method of claim 1 , further comprising utilizing the predicted traffic control feature to validate or modify traffic elements in a digital map.
10. A non-transitory computer readable medium comprising instructions that, when executed by at least one processor, cause a computer device to:
generate a training telemetry digital image comprising pixels representing telemetry information for a traffic area;
determine a predicted traffic control feature associated with the traffic area by processing the pixels representing the telemetry information from the training telemetry digital image utilizing a machine learning model; and
train the machine learning model by comparing the predicted traffic control feature to a ground truth traffic control feature associated with the traffic area.
11. The non-transitory computer readable medium of claim 10 , further comprising instructions that, when executed by the at least one processor, cause the computer device to train the machine learning model by determining a measure of loss between the predicted traffic control feature and the ground truth traffic control feature utilizing a loss function.
12. The non-transitory computer readable medium of claim 11 , wherein the machine learning model comprises a convolutional neural network and further comprising instructions that, when executed by the at least one processor, cause the computer device to train the machine learning model by modifying parameters of the convolutional neural network from the loss function utilizing back propagation.
13. The non-transitory computer readable medium of claim 10 , further comprising instructions that, when executed by the at least one processor, cause the computer device to:
generate an additional training telemetry digital image comprising pixels representing additional telemetry information for an additional traffic area; and
train the machine learning model by comparing, from the pixels of the additional training telemetry digital image, an additional predicted traffic control feature to an additional ground truth traffic control feature associated with the additional traffic area.
14. The non-transitory computer readable medium of claim 10 , further comprising instructions that, when executed by the at least one processor, cause the computer device to determine the training telemetry digital image by determining speeds and locations of transportation vehicles as the transportation vehicles pass through the traffic area.
15. The non-transitory computer readable medium of claim 10 , further comprising instructions that, when executed by the at least one processor, cause the computer device to determine the training telemetry digital image by:
identifying a plurality of speed-location combinations indicating speeds of transportation vehicles at corresponding locations within the traffic area;
determining frequencies with which the plurality of speed-location combinations occur within the traffic area; and
generating pixels of the training telemetry digital image from the frequencies of the plurality of speed-location combinations.
16. A system comprising:
at least one processor; and
a non-transitory computer readable medium comprising instructions that, when executed by the at least one processor, cause the system to:
generate a training telemetry digital image comprising pixels representing telemetry information for a traffic area;
determine a predicted traffic control feature associated with the traffic area by processing the pixels representing the telemetry information from the training telemetry digital image utilizing a machine learning model; and
train the machine learning model by comparing the predicted traffic control feature to a ground truth traffic control feature associated with the traffic area.
17. The system of claim 16 , further comprising instructions that, when executed by the at least one processor, cause the system to train the machine learning model by determining a measure of loss between the predicted traffic control feature and the ground truth traffic control feature utilizing a loss function.
18. The system of claim 17 , wherein the machine learning model comprises a convolutional neural network and further comprising instructions that, when executed by the at least one processor, cause the system to train the machine learning model by modifying parameters of the convolutional neural network from the loss function utilizing back propagation.
19. The system of claim 16 , further comprising instructions that, when executed by the at least one processor, cause the system to determine the training telemetry digital image by determining speeds and locations of transportation vehicles as the transportation vehicles pass through the traffic area.
20. The system of claim 16 , further comprising instructions that, when executed by the at least one processor, cause the system to determine the training telemetry digital image by:
identifying a plurality of speed-location combinations indicating speeds of transportation vehicles at corresponding locations within the traffic area;
determining frequencies with which the plurality of speed-location combinations occur within the traffic area; and
generating pixels of the training telemetry digital image from the frequencies of the plurality of speed-location combinations.