MobilEye has made a bold business announcement. Their system, now named “MobilEye Drive” and consisting of their EyeQ chips and an array of cameras, multiple LIDARs, their REM crowdsourced maps and their “RSS” avoid-illegal-moves motion planning approach will be deployed in shuttles in Israel next year (as previously forecast) and in delivery vehicles in 2023 through a newly announced partnership with Udelv. In addition, they have pending partnerships in several other countries.
MobilEye will use their LIDARs in the radically different way I described earlier, attempting to build two independent perception systems and hoping to fail only when both make a mistake.
Udelv has been working on robotic delivery for some time, and their specialty has been a van with a special package locker inside able to be reconfigured to hold individual parcels and let customers get access to only that parcel. That is probably not that important — who is going to steal somebody else’s package when on camera at their own house? — but an announcement to actually deploy is more meaningful. Udelv hopes to compete with Nuro, which has received at $5B valuation for their 2-compartment delivery robots.
The real merit of the news is that they are nailing down specific dates. It’s very hard to compare the progress of different self-driving teams. Everybody can produce a nice video of driving without interventions — indeed, if you can’t easily make a multi-hour video like that, you aren’t even really a player, but there is a lot of distance to go from such a video.
To really compare, you can either get the chance to look deep inside (which most companies won’t) or the company has to be highly transparent with tons of operations data and safety reports (which only Waymo has been willing to do.) And of course you can deploy real service, which only Waymo and AutoX have done without safety drivers, though a few have operated services with safety drivers on board.
That leaves one last step — what do they have the guts to truly promise, for a specific date, at scale. It’s not a great metric, because promises are not an operating service, but with so few operating services we don’t get a lot of choice.
The announcement by Cruise in a tweet that they will deploy in Dubai with the Cruise Origin shuttle in 2023 has a bit less detail — just the tweet. It says they will be the exclusive provider, yet MobilEye claims they have a partnership announcement pending in the UAE which could contradict that, or be a different Emirate. (The recent opening up of relations between the UAE and Israel has generated a lot of excitement and business opportunities.)