"Until these problems are solved, human beings will remain an integral part of the driving task, monitoring the AI system as it performs anywhere from just over 0% to just under 100% of the driving. The governing objectives of the MIT Autonomous Vehicle Technology (MIT-AVT) study are to (1) undertake large-scale real-world driving data collection, and (2) gain a holistic understanding of how human beings interact with vehicle automation technology."
(1) So much for driverless cars, I guess. (2) I guess a study is needed when aerospace and rail has, oh, 50+ years or so of cumulated data about the problem.
MIT or not, these people seem pretty confused about the rules governing safety in automation, and the abstract is just inaccurate. The driving task can be formalized. It has been done before. [2] Maybe it's a boring driver, but it's possible. Part of this formalization requires the safety control functions to be separated from the driving and navigation functions. AFAIK this principle was formulated first by Georges Westinghouse when he patented his air brake system in 1873 [1].
They've drawn these conclusions based on studies of automated (not autonomous) systems from Volvo, Tesla and Range Rover. They didn't study a single product from a company working on an actual autonomous driving system, and of course they can't because those products are all currently in development and not available to study.
If we built the roads better, such that they facilitate the automation, we'd get there sooner.
For example, RFIDs in the road or on the road side that tell the cars (or actually, allow them to look it up in a database) the exact layout of the road around them.
I think from a technological progression point of view, it's generally wise to adjust the technology to the domain rather than try to adjust the domain to the technology.
For instance what SpaceX is trying to do with retropulsive landings. If they can master this in a variety of conditions then the technology can be adapted outwards to no end. On the other hand, if you focus exclusively on landing back on Earth then landing on another planet is suddenly a huge ordeal. So in this domain, if we master the car driving without too many cues, then a self driving vehicle ought be able to drive in Timbuktu as well as LA, which I think is much more desirable - even if it's vastly more challenging.
far fetched, musk's tunnel idea is not so boring. ultimately you will need to use your car only for last 1 km. rest will be on rail road / tunnel network / rfid network. ( this will play havoc to the automotive industry...that depends on user changing car every 5 to 10 year. )
Far fetched? My jaw drops. It's a moving block signalling system with 1 car trains. This stuff has been in service for decades pretty much all around the world. As for the very concept of the boring company, Elon Musk has it completely wrong. It's not the tunnel that takes so much time to dig, it's the vertical shafts for the stations, especially in the city, because of the things that have to be moved aside, and also because of the precautions you must take to, err, avoid collapsing buildings and stuff. Really, there are things where he knows what he is talking about, but mass transit is not one of them.
I'm still one of those people who really wanted a segway-city. I still have hopes people will build new cities where there is currently no infrastructure. Maybe under there, people can build those hyperloop things.
From the video, they have 8,000 events recorded where the driver took over from Tesla's autopilot. But those include every time someone took over to make a turn or lane change. The interesting ones are when the driver had to take over unexpectedly. Those need to be extracted and studied thoroughly. The speaker says nothing about that.
"Until these problems are solved, human beings will remain an integral part of the driving task, monitoring the AI system as it performs anywhere from just over 0% to just under 100% of the driving. The governing objectives of the MIT Autonomous Vehicle Technology (MIT-AVT) study are to (1) undertake large-scale real-world driving data collection, and (2) gain a holistic understanding of how human beings interact with vehicle automation technology."