I watched the introductory lecture of the MIT Self-Driving Cars series
and these are the highlights of the lecture.
GitHub Tutorials: https://github.com/lexfridman/mit-deep-learning
Waymo has done incredible work in deploying and testing their autonomous vehicles.
The other incredible accomplishment is on the side of the semi-autonomous driving of autopilot. The autopilot miles are driven with the primary sensor being a camera. It’s probably the largest deployment of neural networks in the world that has direct impact on the human life.
On the sobering side, there has been three fatalities.
There has been a lot of announcements of public testing of autonomous taxi services. These companies have been delivering real people from one location to another on real roads. Nuro is doing zero occupancy delivery of groceries autonomously.
Companies have made predictions when they are going to be able to deploy autonomous vehicles.
Many of the leading engineering behind these teams inject a little bit of caution about how difficult it is to remove the human from the loop of automation.
For Elon the autonomous vehicles will be here in 2019. Sometimes you have to believe the impossible is possible in order to make it happen.
Autonomous vehicles will not be adopted because of safety, even though they may be safer. Or because they get us to the location faster, they are going to be slower until the majority of the fleet is autonomous. They are cautious and therefore slower.
For every promise and hope that they will be cheaper, there is significant investment going into them and it is not good economics in the near term to make them significantly cheaper.Autonomous vehicles will be adopted by human beings if they create a better human experience: interaction with the car, natural language communication, ability for the car to sense what you are doing, being able to transfer control back and forth & it is all personalized to you.There are several levels of autonomy. Level 2 is when the driver is still responsible and in 3-5 there are part of the liability that are on the car.
There are only really two levels: human-centered autonomy where the human is responsible & full autonomy where the car is responsible. Both on the legal side, the experience side and the algorithm side.
Full autonomy does not allow for the human to step in and control the vehicle because that means that the human is still in the loop.How do we envision getting this deployed in the mass scale?
Wow, this is huge! There is this idea of connected vehicles: vehicle-to-vehicle communication and vehicle-to-infrastructure communication. enabling us to navigate in an intersection efficiently without stopping and removing all traffic lights.
Layers of tunnels under the cities, constraining the problem to such a degree that the the idea of autonomy is transformed. Lex calls this the boring solution. 😂
And of course there is this idea of personalized flying cars vehicles. One of the companies seriously working on this is Uber, where you meet your vehicle on the roof of a building.
“Many of the great solutions to world’s problems have been laughed at at some point so let’s not laugh too loud at these possibilities. Back in my day we used to drive on the street.” 😂
Poll about who will be the first to deploy 10k+ autonomous cars on public roads without a safety driver shows the current leaders today but it is wide open to ideas. Some companies are taking advantage of the hype and others are really just trying to solve the problem.
Currently Tesla starting with semi-automated vehicles and working towards fully autonomous vehicles and Waymo starting with fully autonomous vehicles are the leaders in the space.