Ride-hailing company Lyft has passed an important mile-marker in its self-driving car program by successfully delivering more than 50,000 rides, making it the most well-trafficked U.S. program of its kind.
According to the company, the milestone marks a ten-fold increase over its total in August 2018 when the program announced that it reached 5,000 rides.
The program currently operates 30 modified BMW’s within a 20 square-mile area in Las Vegas where it delivers riders to some of the city’s most heavily trodden areas, like the Las Vegas strip.
Lyft says the precedent is significant not just for the sheer volume of rides, but for the reported quality.
The average ride rating was 4.97 out of 5 stars and 92 percent of those riders reported feeling ‘very’ or ‘extremely’ safe, reports the company.
Safety ratings have long been one of the most important metrics of nascent self-driving programs who have faced a formidable challenge in breaking down riders’ concerns over letting artificial intelligence take the wheel.
In Lyft’s case, however, even the latest glowing safety ratings by its customers may fail to represent a true stamp of approval.
All of the company’s self-driven rides have been accompanied by a human co-pilot capable of intervening if necessary — a presence which may have served to stoke confidence from its passengers.
Lyft, who launched its program using technology by Aptiv, has not publicly stated when or if it will push to take the next step and get fully autonomous vehicles on the road or whether it plans to expand the program outside of Las Vegas.
Lyft’s program may be the most tested, but other competitors like Waymo have also rolled out their own tests, which include a fleet of self-driving freight trucks in Phoenix and its ride-hailing, Waymo One.
Likewise, Uber recently announced a $1 billion investment in bringing its self-driving car to fruition.
Despite the fact that there are more semi-autonomous cars on the road than ever before, full automation continues to be firmly out of reach in part due to the complexities of both vision and perception involved in the process of safelty navigating a vehicle.
According to MIT Technology Review, in the U.S., one passenger dies once in about 1 million hours of aggregate driving. By eliminating fatalities that involve texting or drunk driving, that figure would rise to about on fatality in about 10 million hours.
Currently, states the report, self-driving systems fail to perceive something on the road one time in every tens of thousands of hours making them leagues below the ability of human-operated vehicles.
Another hurdle for the future of autonomous travel will be less contingent on technology and more reliant on the laws that surround it.
As noted by MIT, regulators will be tasked with determining a comprehensive set of guidelines and safety standards that both protect consumers and allow for incremental progress.
For now, company’s will focus on nailing the basics.
Recently, one self-driving company, Cruise, which is backed by General Motors, showed significant progress in a field that most adept human drivers feel at least mostly comfortable with: making left turns.
According to Cruise, their car safely executed 1,400 ‘unprotected’ left turns — meaning no dedicated signal or traffic pattern — in the span of 24 hours.
Though seemingly mundane, left-hand turns that cross oncoming traffic are part maneuver and part social interaction — drivers often have to ‘negotiate’ with their fellow motorists before deciding who drives when.
Even humans are 10 times more likely to get into an accident when making a left-hand turn as reported by the Washington Post.