one of the lead paper authors (jinhua xu) works at midjourney, appears in the video, and comments specifically on how the midjourney approach is the next, significantly better-funded iteration of the paper approach
The Kyoto Railway Museum (https://www.kyotorailwaymuseum.jp/en/) is certainly worth a mention. Tons of well-preserved examples of Japanese rolling stock, including multiple generations of Shinkansen (including the first!). Their technical explanations of every part of the train were also incredibly well done, with lots of examples from parts of real trains. It was one of the most impressive and genuinely educational technical museums I've ever been to.
Having been to both the National Railway Museum in Taipei and the Kyoto Railway Museum and comparing the two, I'd say that the former was particularly strong in areas around train maintenence, whereas the latter had much more content about trains themselves.
> That's lateral grip, which isn't the same at all. [...] There's a lot of clever things you can do to increase lateral grip, such as wheel camber, that don't really apply to purely longitudinal grip, so I'm not sure this is valid.
Camber isn't a magical trick to get more grip; it's a way to restore grip that would otherwise have been lost because of uneven tire loading in a corner. In a straight-line drive situation, the load is already ideal; the contact patch is the maximum size and fairly evenly distributed across the width of the tire.
> There is just no way you can make a road car that's faster than an LMP1 hybrid.
Indeed, it's currently impossible to make an all-electric race car that can compete with an ICE or hybrid race car in general race conditions, mostly because of the limitations of the energy storage. If the goal is just for a road car to beat a hybrid LMP1 (or even F1) car in a drag race though, as is the case here, I think that's much more doable. The ICE is really the weak link there.
Agreed, but they don't need to be. Remember, the magic number here is ~1.4G, for a 1.9s 0-60. The Pilot Sport Cup 2 – a track-friendly R-compound tire used in the webcast car and in the videos – can pull close to that on a skidpad (i.e. less than optimal conditions), meaning the grip is there.
> I think it's straining credulity to believe that an electronic traction control system is going to outperform them to such a huge degree.
Launch control and traction control can make several tenths of seconds of difference, which is critical when you're talking about sub-2s times. Also, traction control can keep the car on the cusp of slip the entire run to 60MPH, which is critical in a car that has a completely flat torque curve and probably enough torque to break the wheels loose at any speed (which is not true for F1 cars).
I also suspect that the Roadster has active damping – another technology disallowed in F1 – meaning that the duration of contact with the road can be maximized. This is important if the road surface isn't glassy-smooth.
> This car also looks to weigh around double what an F1 car will weigh
That doesn't help it at all in cornering, but in a straight line, the increased weight of the car will help it launch even better since it'll increase the traction on the drive wheels (equivalent to downforce at speed).
> Gear changes: F1 gear changes take about 8 milliseconds. A road-going automatic gearbox is definitely not going to beat this.
There's no gearbox to speak of; the wheels are direct-drive. To be fair, this won't contribute significantly to faster 0-60 times, but the gearbox exists to compensate for some less-than-ideal characteristics of an ICE, namely uneven power delivery and physical limitations on peak RPMs. An electric motor has none of these problems.
Traction control has been banned in F1 since 2008, so this severely limits how quickly they can start from a standstill. The tires themselves certainly have enough grip to handle the acceleration; F1 cars routinely hit several lateral Gs in cornering (though with the help of downforce), well exceeding the ~1.4G required to accelerate to 60 in 1.9s.