Business

Lyft to require multi-sensor systems for driverless rides

Lyft CEO David Risher said fully driverless vehicles must use varied, redundant sensors before they can operate on the company’s platform.

Sofia Marchetti

By Sofia Marchetti · World Affairs Correspondent

3 min read

Lyft to require multi-sensor systems for driverless rides
Photo: Fortune

Lyft will require fully driverless vehicles on its ride-hailing platform to use multiple types of sensors, CEO David Risher said in a Fortune commentary. The policy sets a safety bar for autonomous rides as Lyft adds the technology to a network that handles millions of trips, according to Risher.

Risher said Lyft is updating its AV Partner Safety Evaluation Framework to require what he described as a redundant perception system using different sensor types. He said the rule applies to fully driverless vehicles on Lyft, and does not cover driver-assistance systems used by human drivers in their own cars.

The decision reflects Lyft’s current view of autonomous-vehicle safety, Risher wrote. He said the company has reviewed different self-driving systems and concluded that AVs must use a multi-sensor setup before they can be accepted for driverless service on the platform.

Why Lyft is drawing the line at sensor diversity

Risher cited road safety as the reason for the standard. He wrote that about 40,000 people die and 2.4 million are injured on U.S. roads each year, and said crashes are the leading cause of death for teens and young adults.

According to Risher, speeding, alcohol impairment and distracted driving are the largest contributors to those losses. He argued that autonomous vehicles can help reduce road deaths because they are designed to follow speed limits and traffic laws and do not drive while distracted.

Lyft’s concern, Risher wrote, is that different autonomous-vehicle developers have chosen different sensor designs. Some systems combine cameras, radar and LiDAR so one type of sensing can help cover another when conditions interfere with perception, he said.

Risher said single-sensor approaches carry risks tied to the limits of each technology. He wrote that cameras can be affected by glare, fog or blocked lenses; radar can have trouble with stationary objects; and LiDAR can be weakened by heavy precipitation.

Lyft’s position is that redundancy works best when the backup systems do not share the same weak point at the same moment, Risher said. He also said adding sensors is not enough on its own because the software combining the data must perform well.

Policy could change as technology develops

Risher said Lyft is not ruling out single-sensor systems permanently. He wrote that the company will revisit the policy if evidence or outside standards support a change.

  • Lyft may reconsider if safety data shows single-sensor systems perform comparably to established multi-sensor designs, according to Risher.
  • Lyft may also reconsider if a single-sensor driverless system meets a standard from a credible authority, such as the National Highway Traffic Safety Administration, Risher wrote.

Risher said Lyft expects single-sensor systems to improve and potentially help cut costs, speed deployment and support innovation. He said the company plans to keep publishing updates to its standards as autonomous-vehicle technology advances.

The CEO said autonomous vehicles can be highly safe if deployed properly and described them as one part of Lyft’s broader safety work. He pointed to safety tools in the Lyft app and programs such as Lyft Teen as other pieces of that effort.

This story draws on original reporting from Fortune.