Sensors Converge recap: Debate continues over LiDAR's role in autonomous driving

Navigation systems for autonomous vehicles are among the applications for new sensor from
Some are looking forward to LiDAR as the eventual future of autonomous driving, but Tesla is taking a different route. (Pixabay)

To LiDAR or not to LiDAR: That is the question regarding the future of autonomous driving. If you’re Tesla CEO Elon Musk, the answer is a definite no, but many others have embraced LiDAR

Last week’s Sensors Converge featured two speakers on opposite sides of the issue. Anthony Depaolantonio, associate manager of test engineering at Tesla, told Fierce Electronics Editor Matt Hamblen during a stage interview, “I see eye to eye with Elon on that [LiDAR being unnecessary]. We have two eyes. As humans we don’t have Lidar. We can judge distance, we can see things coming, so why can’t we do it with a car? [Teslas with full self-driving capability] have two cameras. In my mind we have the best self-driving capability out there. Right now, I would say--and Tesla even says to drivers, ‘Make sure you stay in control of the vehicle.”

He added, “That’s why we have input sensing to make sure your hands are on the vehicle. It’s for safety. If the driver doesn’t input, then the vehicle will gradually slow.”

However, attendees of Sensors Converge who were surveyed during a session led by Will Tu, senior director, automotive at Xilinx, feel more strongly that LiDAR needs to play a role in the future of autonomous vehicles. In a stage interview with Hamblen after his session, Tu shared the results, indicating that 95% of those surveyed agreed that Level 4 autonomous driving “would need LiDAR or 4K imaging radar.” (Granted, this survey was taken by a portion of attendees at a sensors event where some companies represented are likely to have a stake in the success of LiDAR sensors.)

Level 4 autonomous driving is generally viewed as the level at which vehicles would be capable of stopping themselves and not requiring the intervention of human drivers to take over driving at any point, although the Society of Automotive Engineers has frequently tinkered with and added clarifications to its six-tier Levels of Driving Automation chart. Also, some limitations eventually could be in play for Level 4 autonomous driving, such as limiting it to use only in optimal weather and road conditions, or with other operational domain limitations in place.

“Tesla would say you don’t need LiDAR,” Tu said, echoing Depaolantonio’s statement that computer vision cameras are enough. 

While Tesla has faced a federal probe into crashes of its vehicles, Depaolantonio told Hamblen, “Tesla has proven to have the safest vehicles on the road,” referring to the company’s five-star ratings from New Car Assessment Programs in the U.S. and Europe.

“You see a lot of publicity toward Tesla because they’re Tesla,” he added. “I mean if you look at the amount of internal combustion vehicles that catch on fire in a year, the amount of Teslas is 0.1% of that, but because Tesla is big [it gets more notice].”

Availability and cost

While LiDAR may be popular, it still faces challenges, particularly having to do with the timing of broad availability of Level 4-capable vehicles and potential cost Level 4 could add to those vehicles. While Tesla has been operating a Full Self-Driving beta program for sometime involving thousands of customers, and opened that beta to even more customers, availability of Level 4 vehicles from the broader auto industry could be years away.

Tu’s survey at Sensors Converge also sought attendee input about the timing and cost of Level 4 autonomous driving, with 43% of those surveyed suggesting it would be 2028 to 2030 before Level 4 would be available to consumers. Another 19% of those surveyed said it would take even longer, sometime after 2030.

However, whenever Level 4 does reach the consumers, it could find willing buyers, at least among Sensors Converge attendees. About 33% of those surveyed by Tu said they would be willing to pay $4,000 or more for the Level 4 capabilities, and 29% said they would pay between $2,500 and $4,000.

“I was a little surprised by that,” Tu said. “We are people that like to have things for free. I thought more people would say $1,000 to $2,500. The thing is no one said they didn’t want it [With this survey question, respondents were given a choice to say they didn’t want Level 4 capabilities at all.]” 

Still, eventual consumer reaction to the cost of Level 4 capabilities likely will be shaped by whether this cost is included in the sticker price of a car or presented aside from the purchase cost as an optional subscription. Tesla’s FSD beta uses a subscription model, and other automakers could offer Level 4 under similar models.

Tu said such models represent the future of the automotive industry as more cars come to include chips, software and automated systems whose capabilities can be turned on or off, or installed or uninstalled, through over-the-air updates. “A lot more automakers will start to look at the vehicle like a mobile phone where you have these features, like a driver monitoring system, that not everyone needs that you can turn on or off.”

RELATED: High-performance LiDAR is key to safe autonomy