NTSB: Uber’s AV in fatal crash ‘had software issues’
The US National Transportation Safety Board (NTSB) has found that an Uber autonomous vehicle which killed Elaine Herzberg last year had software flaws.
NTSB released a report which says the Volvo XC60’s autonomous system software classified the pedestrian as an unknown object and determined that an emergency braking manoeuvre was needed to mitigate the collision.
Uber confirmed that emergency braking manoeuvres must be carried out manually and the system is not designed to alert the driver.
Data
November 6, 2019
Read time: 2 mins
The US 5628 National Transportation Safety Board (NTSB) has found that an 8336 Uber autonomous vehicle which %$Linker: 2External<?xml version="1.0" encoding="utf-16"?><dictionary />000link-external killedfalsehttps://www.itsinternational.com/categories/utc/news/uber-ends-self-driving-programme-in-arizona/falsefalse%> Elaine Herzberg last year had software flaws.
NTSB released a %$Linker: 2External<?xml version="1.0" encoding="utf-16"?><dictionary />000link-external reportfalsehttps://dms.ntsb.gov/pubdms/search/hitlist.cfm?docketID=62978&CurrentPage=2&EndRow=30&StartRow=16&order=1&sort=0&TXTSEARCHT=falsefalse%> which says the 609 Volvo XC60’s autonomous system software classified the pedestrian as an unknown object and determined that an emergency braking manoeuvre was needed to mitigate the collision.
Uber confirmed that emergency braking manoeuvres must be carried out manually and the system is not designed to alert the driver.
Data retrieved from the autonomous system revealed that the vehicle operator engaged the steering wheel less than a second before impact while the vehicle was travelling at 39mph.
The NTSB reviewed Uber’s autonomous system cameras which revealed that Herzberg did not look in the direction of the vehicle just before impact.
An inward-facing video showed the vehicle operator glancing down toward the centre of the vehicle several times before the crash. A post-crash interview with NTSB investigators revealed that she was monitoring the autonomous system’s interface.
The NTSB is working with Uber, Volvo Cars and the 6576 Arizona Department of Transportation to compile a complete account of the crash.
Uber is not the only company whose AV trials have led to a fatality. Last year, a Tesla vehicle operating in %$Linker: 2External<?xml version="1.0" encoding="utf-16"?><dictionary />000link-external autopilotfalsehttps://www.itsinternational.com/categories/utc/news/tesla-car-crash-in-california-kills-driver-while-running-on-autopilot/falsefalse%> in California caught fire after crashing into a roadside barrier that killed the driver.
UK company Traffic Technology’s popular non-confrontational community speed watch device, SpeedWatch, is now able to target drivers’ distracted or dangerous driving behaviour, such as mobile phone use or failure of occupants to fasten seatbelts. The SpeedWatch+ interactive display wirelessly interfaces to the manned SpeedWatch radar system, while a tablet device enables trained members of the public to select appropriate messages which are sent to the display to warn violating motorists. A large aesthetic
Google has announced the next step in its autonomous vehicle program and is about to begin testing its new prototype self-driving vehicles on public roads. This summer, the company will move its cars from the test track to the roads with safety drivers aboard.
The company has been rigorously testing the cars at its test facilities for several years. The new prototypes are based on the company’s existing fleet of self-driving Lexus RX450h SUVs, which has logged nearly a million autonomous miles and recen
Evikey has launched a secure USB drive solution controlled from an NFC-enabled device. Self-powered and battery-free, it has an intelligent diagnostic system that enables administrators to define and track how and when the devices can be used and by whom.
Cestel says its MkIII SiWIM technology turns any bridge into a fully automatic weigh-in-motion system without disturbing the road surface. Sensors are fixed to the underside of the bridge deck beneath up to four lanes of traffic and the system is calibrated with vehicles of known weight passing individually, in opposite directions and in various combinations depending on the road layout. Having been calibrated, an algorithm then interprets the information to determine the number, weights, speed and