SoylentNews is people
SoylentNews
Submitted via IRC for Runaway1956
Tesla fatal crash: 'autopilot' mode sped up car before driver killed, report finds
A Tesla driving in "autopilot" mode crashed in March when the vehicle sped up and steered into a concrete barrier, according to a new report on the fatal collision, raising fresh concerns about Elon Musk's technology.
The National Transportation Safety Board (NTSB) said that four seconds before the 23 March crash on a highway in Silicon Valley, which killed Walter Huang, 38, the car stopped following the path of a vehicle in front of it. Three seconds before the impact, it sped up from 62mph to 70.8mph, and the car did not brake or steer away, the NTSB said.
[...] The NTSB report [...] has once again raised serious safety questions about the limits and performance of the autopilot technology, which is meant to assist drivers and has faced growing scrutiny from experts and regulators. Mark Fong, an attorney for Huang's family, also said the report appeared to "contradict Tesla's characterization" of the collision.
The NTSB press release includes this link to the preliminary report, for anyone inclined to read the slightly longer version of events.
The Mountain View Fire Department applied about 200 gallons of water and foam to extinguish the post-crash fire. The battery reignited five days after the crash in an impound lot and was extinguished by the San Mateo Fire Department.
Layoffs at Tesla
Tesla Lays Off 9 Percent Of Workforce
Tesla will lay off about 3,500 workers in an effort to boost profitability, CEO Elon Musk wrote in a company email.
"What drives us is our mission to accelerate the world's transition to sustainable, clean energy, but we will never achieve that mission unless we eventually demonstrate that we can be sustainably profitable," Musk wrote.
Musk conceded that Tesla has not made an annual profit in 15 years. The company posted its largest quarterly loss, of more than $700 million, earlier this year.
Original Submission #1 Original Submission #2 Original Submission #3
(Score: 2) by GreatAuntAnesthesia on Thursday June 14 2018, @04:18PM (3 children)
I don't see why not. 99% of cars are currently driven by rather esoteric software running on squishy neural networking hardware. I don't see why we couldn't do something similar in silicon.
(Score: 2) by tangomargarine on Thursday June 14 2018, @04:36PM (1 child)
Are you a programmer?
"Is that really true?" "I just spent the last hour telling you to think for yourself! Didn't you hear anything I said?"
(Score: 2) by SomeGuy on Thursday June 14 2018, @05:03PM
Clearly he works in management. :P
(Score: 2) by VLM on Thursday June 14 2018, @10:03PM
The real solution is parallel processing. With massive increases in income inequality and inflation driving the cost of a fast electric car ever higher, there will never be anything other than a small very rich minority of people owning and driving electric sports cars.
The legacy population in the USA likes to LARP like they're middle class, even when they aren't, so having servants like a nice driver is not possible... directly... however indirectly, hook up a large odd number of drivers from rural or poor USA to each car, and average their responses... So you get a wisdom of the crowd effect, combined with using AI appropriately, in this case AI would "monday morning quarterback" each driver's extensive historical record of driving output vs sensor input vs crowd decision, to give them a weighting on the control of the car, but never give any driver more than, perhaps, 1/5 authority over the car. Because of massive income inequality you can have maybe 1/2 the drivers at any given time running pure sims for AI training purposes.
Odds are we're going to run out of cheap gasoline before we run out of poor people to drive rich peoples cars, and once we run out of gas guzzlers the market will fall out from under, making road travel roll back to horses and such making the whole topic moot. You don't need a solution that scales to seven billion middle class (or higher) people for centuries... you merely need to keep a couple rich jackasses puttering about in a "self driving car" until we run out of cheap gas, probably no more than a decade or two at most.
This should be scalable for aircraft.