Harvard/MIT Research on Tesla Driving/Performance (Machine Learning)

[BizJet]

A group of graduate students I know from Harvard and MIT are looking for volunteers to participate in a machine learning / data-mining project designed to give new insights about driving, performance, routing, etc. to drivers of Tesla automobiles.

They're seeking Tesla drivers who are willing to grant API data access to their cars for a period of time, in order to collect available API data for their project. They, in turn, promise to share what they learn with the participants.

I personally know the leader of this group, and he is of the highest integrity. And the principles they've put forth for data protection are solid.

I'm participating, and would urge fellow Tesla owners who are interested in getting some fresh insights to check it out.

Here is a link with more info:

Harvard / MIT machine learning research

...and the sign-up is at this link:

teslaresearch

 

[adamjb]

Thanks, Michael for sharing. If anyone has any questions--happy to answer on the forum, or directly via email at adamjb at mit dot edu.

 

[Tiger]

Training methodology for autonomous vehicles

 

[adamjb]

Great link, @Tiger! There's a bunch of cool research happening on the autonomous side of things with computer vision, lidar and other sensors. Our research is in the adjacent department as the deep learning research--in the Brains, Minds, Machines department which is approaching the problem of cognition from a probabilistic and semantic computational angle. And for this particular project, we're only using the constrained set of data exposed via the car API.

We hope the two feed into each other, e.g. learn from driver behavior to actually improve the 'higher-order, conceptual thinking' of AI systems in the vehicle.