Researchers at McAfee adding a small piece of tape to the "3" on a 35 MPH traffic sign, to make it look a little like an "8." Then they put two older model Teslas in autonomous mode, and both accelerated to the new speed limit.
Don't try this in a school zone, kids.
From Business Insider:
Tesla's newer models use proprietary cameras, and MobilEye EyeQ3 has released newer versions of its cameras that McAfee tested and said were not fooled by the modified sign.
The McAfee researcher Povolny told MIT Tech Review that the findings were still concerning, though, as plenty of 2016 Teslas are still on the roads. "We are not trying to spread fear and say that if you drive this car, it will accelerate into through a barrier, or to sensationalize it," he said. "The reason we are doing this research is we're really trying to raise awareness for both consumers and vendors of the types of flaws that are possible."
Image: McAfee Read the rest
From The Tesla Show Twitter feed: "What unreleased FSD Autopilot sees. Straight from Tesla Autopilot recruiting website."
From the job description:
Apply cutting-edge research to train deep neural networks on problems ranging from perception to control. Our per-camera networks analyze raw images to perform semantic segmentation, object detection and monocular depth estimation. Our birds-eye-view networks take video from all cameras to output the road layout, static infrastructure and 3D objects directly in the top-down view. Our networks learn from the most complicated and diverse scenarios in the world, iteratively sourced from our fleet of nearly 1M vehicles in real time. A full build of Autopilot neural networks involves 48 networks that take 70,000 GPU hours to train ?. Together, they output 1,000 distinct tensors (predictions) at each timestep.
Image: Twitter screengrab Read the rest
This seems like some impressive trolley problem decision-making.
Image: YouTube Read the rest
Someone in another car spotted this couple sleeping in a self-driving Tesla that was zipping along the Massachusetts Turnpike, according to NBC on YouTube. The person honked, trying to wake the driver up, but "it didn't work at all." Read the rest
In MobilBye: Attacking ADAS with Camera Spoofing, a group of Ben Gurion security researchers describe how they were able to defeat a Renault Captur's "Level 0" autopilot (Level 0 systems advise human drivers but do not directly operate cars) by following them with drones that projected images of fake roadsigns for a 100ms instant -- too short for human perception, but long enough for the autopilot's sensors.
Read the rest
Bruce Sterling republishes the acronyms in a recent Daimler white-paper on self-driving cars:
Read the rest
The company says 190 employees in Santa Clara and Sunnyvale, CA will lose their jobs.
People like this guy waving his gun at a driverless Waymo van in Arizona are attacking self-driving vehicles with rocks, knives, and *their own cars*, sending a message to tech companies like Waymo, which is owned by Alphabet (Google's parent company). That message is, please go experiment with artificial intelligence in somebody else’s neighborhood. Read the rest
This week, a self-driving Uber killed a pedestrian in Arizona, the first pedestrian fatality involving an autonomous vehicle; in his analysis of the event, Charlie Stross notes that Arizona's laws treat corporations that kill people with considerably more forbearance than humans who do so, and proposes that in the near future, every self-driving car will be owned by a special-purpose corporation that insulates its owner from liability.
Read the rest
A trio of engineering researchers from UC Berkeley modeled the effect of heavy reliance on GPS routing on municipal road efficiencies and found that people who are GPS-routed are likely to move to surface streets and secondary highways when the main highways are congested; though this increases overall throughput in a city and reduces overall drive-time, it also creates heavy traffic on residential streets, effectively transferring traffic jams from highways to neighborhoods.
Read the rest
Self-driving cars require incredibly accurate, up-to-date maps; in China, only Chinese companies will be able to make these maps. Nominally, this is about preventing espionage, but it also has the non-coincedental effect of forcing foreign autonomous vehicle companies to partner with (much more easily controlled) Chinese firms, a policy already in place for traditional auto manufacturing.
Read the rest
Tim O'Reilly (previously) is my kind of technologist: someone who goes past the "is technology good or bad for us?" question and dives into the really meaty, important question, namely: "how can we make technology better for us?" Read the rest
Alexis Madrigal got a chance to visit the fascinating town of Castle, a roads-only city constructed by Waymo for the sole purpose of developing self-driving cars. Read the rest
We already know Uber's been investing millions of dollars in the future of self-driving. Now Lyft is making similar moves, including a partnership with Boston-based nuTonomy, a self-driving car startup founded by an MIT guy. Read the rest
The Chinese tech firm Baidu said Tuesday it will launch a self-driving car technology in July. A first release will involve a restricted environment, but the company plans to gradually introduce “fully autonomous driving capabilities on highways and open city roads by 2020,” Reuters reports.
Read the rest
Charlie Miller made headlines in 2015 as part of the team that showed it was possible to remote-drive a Jeep Cherokee over the internet, triggering a 1.4 million vehicle recall; now, he's just quit a job at Uber where he was working on security for future self-driving taxis, and he's not optimistic about the future of this important task. Read the rest
Any well-designed self-driving car will be at pains to avoid killing people, if only to prevent paperwork delays when they mow someone down. Read the rest
