Crowd-source Driver Input to FSD


Now Implemented!


Crowd-source EAP/FSD raw inputs to Tesla.


The slow progress with EAP and the long wait for FSD is frustrating for many. Getting to Elon’s objective of 10 times safer than a human driver may take years. I suspect there are many enthusiasts who would love to contribute their effort and talent to the project. Perhaps Tesla could set up a ‘Common Task Framework’ via GitHub or Azure ML

If they provide a way to download a recording of all raw inputs (and geolocation plus timestamps) over a short time interval, crowd-sourcing could help get it done faster. The recording could be manually triggered in a situation that is not handled well by EAP, and later uploaded via the owner’s WiFi to a dedicated server. Recording and uploading could be controlled through the Tesla App. Only the owner could download or grant access to the recording.

Tesla would provide details of the data structure as recorded, format for submitting proposed improvements, and perhaps define tasks (including selected user suggestions), specifications (and rewards?). Incentives could include public recognition, a job interview, free supercharging, free service, free EAP / FSD, a car.

Moderator: Keep in mind getting data is only a very small part of the development, likely under 1% of the effort. The hard work is processing the data into test cases, and writing the software to handle various scenarios without affecting other prior work.


Tesla has stated they used some crowd-sourced data for narrow and unusual cases.

lightly edited by moderator
Category: Y3XS Applies to:
     Created 9-Oct-2017