OEMs' L4 autonomy interest increasing despite scale challenges

Questions have abounded throughout the automotive industry — ever since EV market leader Tesla put so many of its eggs in the Level 4 autonomy basket with its robotaxi ploy — as to whether high level autonomous driving (AD) is too risky a bet on which to stake a world-leading automaker.

Tesla's insistence on L4 and above AD as the future of the company makes it an outlier among prevailing OEM strategy trends, especially outside China. For example, German players BMW and Mercedes are among just a handful that have stuck their neck out for even L3 — and the associated legal liability that comes with it.

French OEM Renault Group recently announced its intention to put off any plans for high-level autonomy in passenger vehicles and instead reserve L4 for limited-route shuttle buses.

Elad Hofstetter, chief business officer of Israeli lidar company Innoviz, assures EV inFocus that interest in L4 has not diminished amongst major automotive players. But the biggest roadblock for the technology is difficulty in reaching scale when compared to L3 AD.

"The whole [L4] system is more expensive — the volumes are lower. And the problem when the volumes are lower, you do not have the scale, and then it is harder to drive to get a better cost," Hofstetter says.

"While for L3 there are less than 100mn new vehicles on the road each year, but there are still existing vehicles on the road, so higher volumes," he continues. "All you need to do is just put a sensor inside. The market is already there — you just need to give a push here on autonomy and then have the scale."

Tesla does not use lidar for its autonomy, instead preferring a suite of cameras and radars, but it is the only company pursuing high-level AD without it.

Innoviz is a leading player in L3 projects across major automakers, currently providing its Innovis One lidar sensor and its perception stack to BMW for an L3 programme on the firm's vehicles. Innnoviz is also working on both an L3 and an L4 programme with fellow German heavyweight VW, and undergoing strenuous testing of its products and software with these and other OEM partners.

Incremental approach

Despite much talk around a future full of robotaxis, Hofstetter tells EV inFocus that, as Innoviz scopes out demand for AD amongst automakers, "in terms of volumes, we see the majority of volume going into L3 coming from the OEMs".

This focus on L3 indicates that the industry is being largely patient with its autonomy efforts, preferring to shoot for the next incremental step in private car autonomy features: enabling users to take their hands off the wheel during highway travel.

This is not the grand ambition of robotaxi services, which Tesla has recently put front and centre of its business. But could an L3 focus could risk being at the expense achieving L4 in cases of simple, predictable routes not on public roads —such as commercial space logistics in warehouses or ports — similar to Renault's more limited vision?

Renault characterises its approach as "a pragmatic vision of technology" based on the current limitations and challenges around AD tech, which the firm believes makes public transport the sensible first use case to target — where manageable, repetitive routes present nothing like the challenges of L4 or L5 robotaxis.

As for L4, Hofstetter remaind quietly confident. The eventual step-up to widespread robotaxi-esque capabilities will, in his view, be be a result of simply more time for testing in more and more environments, rather than the result of large leaps in technology that do not yet exist. There is also, he says, "a lot of here of leverage and synergies between the programmes as there are many things in common, both for us supporting them, but also for the OEM".

"I think that there will be more leaps on the L4. I am not saying there will not be. They are less cost-driven in terms of the pricing of the sensors and the sensor suite, not related just to lidar," Hofstetter continues.

Under the radar

While Hofstetter acknowledges that the majority of automakers' focus is on L3 — largely a consequence of the cost structures which allow L3 to reach economies of scale — this does not in itself mean that volumes are down on L4 hardware orders.

"Around 2018-19, there was this hype cycle around autonomy, when everyone felt that autonomy is just around the corner and here it is coming, not just related to lidar but related to the whole industry," Hofstetter says.

But, from Innoviz' point of view, while L3 is taking the spotlight at the moment, ambition for L4 is bubbling away in the background.

"What we see today, or in the last maybe couple of years, in terms of [L4] volumes and real programmes, they are exceeding the volumes we have seen during the hype cycles. Volumes coming and requests coming from OEMs, but less hype," Hofstetter says.