Drivers are not ready to give over the wheel even when +37,000 people died in motor vehicle accidents on US highways in 2016 —most of the cases due to some kind of human error. There is a concern (or fear?) if autonomous cars and trucks are ready to leave the test roads and start driving wide-scale together with human traffic: Are autonomous vehicles safer than human drivers?
Self-driving vehicles are indeed not perfect. Several car makers, technology companies, and startups testing autonomous cars in California are submitting “disengagement reports” to rank the reliability of their driverless vehicles as proof their cars and trucks are getting better at operating with only occasional human intervention; today, autonomous vehicles are almost reliable as a standard crappy driver. However, even when engineers might not mind less-than-perfect driverless vehicles around the streets, causing fewer deaths overall than human drivers (would be a win if self-driving cars get 37,460 deaths instead of the 37,461 caused by humans), the general public is having second thoughts after the death of a driver on a Tesla car —despite thousands have been "saved” by non-distracted, non-emotionally compromised, non-drunk autonomous vehicles.
Will be a moment where driverless cars and trucks own the road? If so, when? The Enemy of Good - Estimating the Cost of Waiting for Nearly Perfect Automated Vehicles research from RAND global policy think tank set intellectual rigor into the question. The researchers examined three basic scenarios using a"robust decision making" analytic method for autonomous cars and trucks driving on the roads:
- Vehicles being 10% safer than the average human driver.
- Vehicles being 75% safer.
- Vehicles being nearly perfect: 90% safer.
RAND clichéd but meaningful outcome is: do not let the perfect be the enemy of the good. They found that driverless cars and trucks could save around 3,000 lives a year while being just a bit better: the 10% safer scenario. Researchers concluded that dozens of thousands of lives could be saved by self-driving vehicles even if regulators allow deploying less-than-perfect autonomous vehicles on the road: letting driverless algorithms becoming better while performing in the real world, making all necessary hardware adjustments to keep upgrading, and taking advantage of the ability to improve as a fleet —when one vehicle makes a particular mistake, all others can learn to avoid it.
IMAGE: Why waiting for perfect autonomous vehicles may cost lives - RAND
Even when the US Department of Transportation identified human errors caused 94% of fatal crashes, the Americans’ attitudes toward driverless vehicle survey by Pew Research Center got 56% of Americans would not ride in a driverless vehicle if given the opportunity. So policymakers and government must keep updating stringent guidelines, frameworks for regulations, and policies around autonomous vehicles while demanding for studies and performance reports, as numbers-based studies could help to modulate public fear when an autonomous vehicle inevitably fatal crashes, tempering the overextended emotional backlash.
Both safety of the roads and lives to save in the future will rely on robust plans created today together with a good public number-based understanding of the risks, while engineers improve the less-than-perfect autonomous technologies to build reliable driverless cars and trucks.
Top Comments