Search results

If there were e.g. a propellant depot on Deimos, a returning Starship could have enough extra delta-v to bleed off quite a bit of speed (~4km/s) approaching Earth entry, at the expense of transit time. But by the time this is happening regularly, ion propulsion may be a feasible and better...

Agreed, which is why connecting the tiles to the ship via physical interlocking may be a more robust solution than adhesive. If the ceramic were physically interlocked with e.g. a carbon-carbon skeleton, and the skeleton were physically interlocked with the ship’s stainless frame (perhaps via...

Does this imply that Starship’s reusable tile approach won’t work for Mars and Mars return? Could they simply make the tiles thicker to compensate for escape-velocity reentry heating? Re tiles falling off, is it known by what precise mechanism they tend to detach? Is it due to the brittleness...

Relative speed of spacecraft and atmosphere. The atmosphere spins along with the planet, so technically that should be taken into account. (E.g. there would be more heating when reentering from a high-inclination or retrograde orbit for this reason.) But to a first approximation, “absolute...

Part of the reason (I think) is that methane is really not the best liquid to use for transpiration. It pyrolizes in high temperatures, causing coking (similar to burning "fuel-rich") which could clog up the transpiration pores. LOX is not much better; pure oxygen is highly reactive. It's...

Brain fart, I meant HRSI ceramic. My point was that if connecting the current tile material directly to steel seems unreliable (as evidenced by tiles continuing to break/fall off), then tapering the material from HRSI on the outside to something tougher on the inside (perhaps via 3D-printing)...

Check your calculations? With Raptor v3’s anticipated 382s vacuum ISP, and 200T final mass (including landing fuel + reserve), I get ~408T of propellant needed to decelerate by 4166m/s. So a bit less than half a tank, but still a lot. Attaching the tiles directly to each other (via wires at...

Starship would need to have a nearly full tank (or at least 1/2 tank) to accomplish this, far in excess of its nominal payload capacity. This would require several extra Tanker flights and an orbital refueling stop per mission, but note that Tankers must also be able to reenter safely, without...

Temperature is determined by relative speed. Decelerating more slowly may counterintuitively mean a higher temperature for a longer period, though counterbalanced by a lower pressure. Effective heat flux is maybe a better metric. A secondary heatshield layer of e.g. honeycomb alumina underneath...

Ludicrous, Insane, Plaid... all they need is a 'Maximum' trim, and they'll be LIMP. Also, their full-self-driving software was first Beta, then Supervised... that's some BS right there. 🙃

Projections, not "expectations". See page 73: https://courts.delaware.gov/Opinions/Download.aspx?id=359340 At the time of the negotiations in Dec 2017, Tesla's internal projections already showed them reaching 11 of the 16 operational milestones by 2020, even without further involvement by...

If Robotaxi is a much different form factor (smaller) than S3XY, it would presumably require its own specific network, similar to how Cybertruck apparently requires it. Also, if the Robotaxi sensor suite is significantly different from S3XYC (e.g. +radar, +lidar, +ultrasonics, +more/different...

Chemistry-wise, I'm not sure what happens to orbital-velocity methane as it reenters. (Thermodynamically, LEO speed is roughly equivalent to the gas being heated to ~80,000C.) It may spontaneously disintegrate into carbon and atomic/molecular hydrogen when it hits another molecule traveling that...

He prefers FoxNewses.

I’d like to see ED pills marketed this way… “If it’s half-up, it’s good to go.”

In 2017 Elon said SpaceX would fly private citizens around the Moon in 2018. He is consistently about 10x too optimistic on short-to-medium-term timeframes.

Considering how much glare/interference there already is between repeater cameras and turn signal blinkers, I wonder how they would avoid the glare problem at night if the bumper cameras are literally inside the much brighter headlight assembly. Or in rainy conditions, brightly lit (by the...

They'd be squashed, at any rate.

The car does not learn on the fly, at least not in the sense of updating its own neural network. There are several reasons for this. First: learning (i.e. "training", for a neural network like FSD) is INCREDIBLY expensive. It takes gigawatt-hours of compute to produce each incremental iteration...

Prediction: 12.4 will have it, but limited to driving in the forward direction only. 12.5 will be ASS backwards.

Their cars may be S3XY, but their autonomy is βS. [STU, it’s a joke.]

I had a very similar situation on the freeway a few months ago (driving manually). Hit some unexpected flooding in the roadway, and had zero visibility for a few seconds; Braked hard and steered by dead reckoning until the windshield cleared. FSD would surely have panicked. Yet an L4 car will...

Spring + latch sounds reasonable, in vacuum. I expect they might do this soon after boostback shutdown, or perhaps even during the flip maneuver to orient for reentry, using the centripetal spin to detach the ring, similar to the IFT-1 stage separation maneuver, although we all know how well...

Airflow from below should compress the hot staging ring against the stack, unless the ring somehow caused so much drag that it’s effectively acting as a parachute. (In which case it would reduce the stack’s terminal velocity, not increase it.) I don’t think this is plausible for the current...

Understood, but they should be able to easily make up for that by reducing Starship payload in these early flights. Unless the v2/v3 hot-staging design is expected to be so different that there’s no useful information they could gain from recovering this version? And how much extra...

I’m surprised they’re doing this; I would have assumed they would want to recover and inspect the hot-stage adapter once they’re able to land boosters. Unless they’re planning to recover them fairing-style under a parachute, though I can’t imagine they float very well? And since it only affects...

This one time, on Hwy 1...

For situations like a dogleg turn, it has to be a hybrid approach. The right thing for the neural network to do at any given moment depends on both the video input and on the navigation goals (particularly if the navigation goals involve an upcoming turn or merge), and so the network will have...

It does have "object permanence", which cuts both ways. If it sees a car passing behind another car, it will project/extrapolate its path even while its view is blocked. The flip side is that if it sees an empty patch of street (as it does around 29:12 in your video), it will assume (on the...

There’s an interesting variation on the trolley problem where the two [and only two] options are to hit a bicyclist who’s wearing a helmet, or to hit a bicyclist who’s not wearing a helmet. By the above logic the car should hit the helmet-wearing bicyclist [since they’re more likely to survive]...

My 2022 Model Y (fully under warranty) just started throwing this alert. I have service scheduled for next week, and will keep this thread posted how it goes.

As noted, one of the advantages of having higher-resolution cameras is that they can serve redundantly for the narrower-FOV cameras, and that's probably how Tesla was able to reduce from three to two physical front cameras from HW3 to HW4. What's more, it could allow the simulated narrow-FOV...

For a low-frequency launch site, using the same tower for launch and landing may make sense. (Although it still may be a good idea to have a second tower for redundancy in case something happens to the main tower.) But for a high-frequency launch site, such as launching 8000 tankers to get 1000...

Pointy end up, flamey end down?

I imagine they would implement it to make the vehicle dynamics precisely mimic a very-high-grip tire on a dry track surface, even when the road is wet. If they could use the thrusters to e.g. simulate tires with 1.5 g lateral grip, that could be intuitive enough for high-performance track use...

Imperial dozens or metric dozens? [Edit] Ah. The royal 'dozens'.

If the FSD network is this sensitive to resampling noise, that's a very bad sign, given how robust to all sorts of other environmental noise it needs to be. (Dust, dirt, rain, condensation, pigeons, etc.) My guess is that the downsampled HW4 images will generally be far higher quality than...

Not per se, but it still does a pretty terrible job with U-turns. There's a spot in my area where the Nav always suggests the U-turn, and FSD always blows straight past it and reroutes. And another where FSD slows to a complete stop in traffic before making a left turn into a protected median...

I dont think both stacks are ever running simultaneously, except maybe briefly during handoffs. But in general, one stack should not slow down the other. The question is whether the eventual combined E2E network (v12.5) may have to be larger or have a different architecture than the current...

I Want My Mommy Average Balls to the Wall.

A heavy abstraction cost is usually incurred when the underlying hardware instruction set has to be emulated, but I doubt this is the case for HW4 vs HW3. More likely HW4 can natively run all the HW3 code and models, perhaps with stalls added to simulate HW3 latency, leaving much of its compute...

I've never gotten used to the on-screen side-view mirror feeds when the blinkers are on, though this is mostly because the car is moving and it's safer to keep eyes on the road. But bumper-cam views in non-moving blocked-visibility situations may be useful enough to overcome this. The canonical...

My understanding is that the Falcon 9 launch explosion was due to an issue specific to the composite fuel tanks, which is a failure mode that may not apply to Starship. To the extent the Starship / SH fueling process is "safer" than Falcon 9, it may be a good enough compromise to chill the...

I'm expecting the eventual E2E Highway stack will greatly improve HOV behavior, so I'm ok with giving the v11 stack a pass til then. (Agreed it needs a lot of work.) What I would really like to see though is "Adaptive" HOV usage, where the car uses the existing seat sensors (and/or cabin...

Elon is not even playing zero-dimensional chess. Then he would at least have a point.

For a launch pad abort, it would probably need the thrust of all nine Raptors. (Assuming Crew Starship will use the 9-Raptor configuration.) I’m not sure what happens when vacuum Raptors are fired at full power at sea level; maybe part of their bells could detach in such a scenario to make them...

The default for any given video clip is not to be used in training; the overwhelming majority are not. 99.9% of fleet-captured video is never sent to the mothership. The remaining 0.1% is sent either on selective request (if Tesla requests that the fleet upload videos of a particular unusual...

I suppose the car could get around this by moving slowly at first to let the driver see that the coast is clear, then completing the maneuver. But the more important reason is to reduce cases where the car thinks the coast is clear, but is wrong. E.g. if the 8-camera setup can see just far...

How is fault proven for non-AV collisions? There is no ground truth in those cases either; the best case is that there's dashcam video, which may be sufficient to give an approximate idea of what happened and who is at fault. If Tesla ever achieves a pure-vision Robotaxi, its eight cameras...

This describes a classical database lookup, not a neural network. A neural network does not maintain a database of its inputs; all the information in the input training set gets irreversibly (and lossily) condensed and mashed together into the neural network, with no way to retrieve or extract...