Training of vehicles to improve autonomous capabilities
Systems and methods to improve performance, reliability and learning to enhance autonomy of vehicles are disclosed. Responses of subjects to non-events and outside events when operating vehicles under various conditions are scored partly based on vehicle location and an associated map. Scoring for non-events is based on performance and mental functioning on selected categories, while scoring for outside events is based on success and failure and mental functioning when responding to these events. Subjects are ranked and expert drivers identified depending on their raw or scaled scores, or a combination score, or a threshold score. Responses of expert drivers are then used as vehicle training data to enable or improve vehicle autonomy. Response data can include data from vehicle sensors, outside environment sensors and human sensors.
1. A method comprising: from a group of one or more subjects each operating a vehicle, scoring each subjects' responses to one or more events occurring outside the vehicle to obtain a total event score for said each subject, and scaling the total event score; scoring said each subjects' performance when there are no events occurring outside the vehicle to obtain a total non-event score for said each subject, and scaling the total non-event score, the scoring of each of the event and non-events at least partly based on a location of the vehicle and an associated map; adding the total or total scaled event score of said each subject to the total or total scaled non-event score of said each subject and thereby obtain a combined score for said each subject; choosing a threshold score; selecting, from the group, the responses of one or more subjects whose combined scores are higher than the threshold score; using the selected responses as vehicle training data, wherein the vehicle training data is used to train vehicles to become partially or fully autonomous or to improve their autonomous functioning.
2. The method of claim 1 , wherein expert drivers are identified from the group by either ranking said each subjects according to their total or total scaled event scores, or according to the combined score of said each subject, or by applying an expert driver selection criteria.
3. The method of claim 1 , wherein the conditions of operation of the vehicles are the same.
4. The method of claim 1 , wherein the conditions of operation of the vehicles are different.
5. The method of claim 1 wherein the subject is a human.
6. The method of claim 1 , wherein the subject is non-human.
7. The method of claim 1 , wherein the vehicle is a real vehicle operated in a real world.
8. The method of claim 1 , wherein the vehicle is a virtual vehicle operated in a virtual world or camera world.