Systems and methods for automatically training neural networks

A method for automatically training a neural network associated with a trainer, the trainer having a first communication device and a perception recorder, the method including:

• at the perception recorder, continuously recording the surroundings in the vicinity of the trainer isolating one or more objects in the recording and attaching positional and timing information to each of the isolated one or more objects, the positional information indicating the location of the corresponding object with respect to the trainer and the timing information indicating a time at which the corresponding object was in that position;
• receiving, at the trainer, a communication message from a communication device associated with a first object in the vicinity of the trainer, the message including positional information about the first object and information about the type of the first object;
• matching a time at which the message is received with the timing information of the isolated one or more objects and matching the positional information received in the message with the positional information of the isolated one or more objects to identify the first object in the recording;
• classifying the identified first object based on the information about the type of the object;
• and
• using the classified recording to train the neural network.

A method for training a neural network, the method including:

• at a vehicle having a first communication device and a vehicle condition recorder, the first communication device configured to receive communication messages from one or more objects in the vicinity of the vehicle and the vehicle condition recorder configured to continuously record a condition of the vehicle;
• receiving, at the vehicle, a first communication message from a communication device associated with an object in the vicinity of the vehicle, the first communication message including positional information about the object and information about the type of the object;
• determining a surrounding state of the vehicle based on the received communication message, the surrounding state including at least a position of the object with respect to the vehicle;
• identifying a vehicle state corresponding to a time at which the communication message is received from the object;
• correlating the determined surrounding state of the vehicle with the corresponding vehicle state and storing in the form of correlated state data;
• classifying the correlated state data;
• and
• using the classified correlated state data to train the neural network.