That's an interesting video, but the commentary is a bunch of Subaru BS. The symmetricality of the system hardly matters in practice. Torque steer only happens when you suddenly load the drivetrain (like flooring the gas), and the left and right shafts react differently. Proper engineering, such as using half shafts with different stiffness (different diameter should do it), could compensate for it if manufacturers thought it was necessary.
2:10 shows the FWD bias. The CR-V, and probably the rest of the vehicles, could easily pass this test. I don't know anything about the Highlander system or how it compares to the 4runner, but I suspect it's not as good as the 4runner. I would guess it's a clutch-based system.
At 7:02 maybe it's the fact that three wheels are on ice, and one wheel in the corner is pushing the car. They suggest that the asymmetricality of the drivetrain is causing the car to rotate. The Subaru doesn't rotate because it gets moving, so the movement of the front wheels corrects the direction of the vehicle. The rollers exaggerate the rotation of the VW. Rollers are a weird surface, and I've biked on rollers so I know.
The 4runner uses a Torsen center diff, similar to Audi Quattro systems. It would be interesting to see that tested. It doesn't use friction, and to be honest I really don't understand how it works. All I know is:
so it might not work well with big speed differences between the front and rear axles. Of course, with the center locked it would easily pass the front or rear wheels on ice test, and it would probably do OK at the three wheels on ice test since it can brake the left or right wheels.
The Honda system is known to be weak in the rear end. And the Highlander system probably is, too, if it shares the system with the RAV4. I know the highlander hybrid AWD had a 68hp rear motor, so that's the most you could get back there. And the motor tended to overheat off road. It's a cool idea to get rid of the mechanical linkage, but not the best implementation.
Honda't system doesn't take control away from the driver (unless traction control is limiting the engine power). It tries to engage the rear wheels, but it can only do so much, and then it backs off the rear clutch to keep from burning it out. In normal conditions it works most of the time, but it is known to be a weak system. It provides enough kick to stop the front wheels spinning, get the car moving, and get traction in the front. Basically it works like the clutch in a manual transmission where it gets you going, but you can't let it slip for more than a second. But it's not strong enough to lock like the clutch in a manual transmission. It's optimized to be lightweight and work on paved roads. Their VTM-4 system in the Pilot and Ridgeline isn't perfect either, but it's a whole lot better. There's a separate clutch for the left rear and right rear wheels, so it can send power to them individually. I don't know how much of the power it can send there.
One could say the traction control system in the highlander takes away control from the driver, but it is doing what it can within its design limits. The all wheel drive system can't handle sending full throttle engine power to the rear wheels so it reduces the throttle and does what it can. Most drivers are idiots, anyway, and floor the gas when their wheels start spinning. I see it every time it snows in Colorado. The highlander knows that won't help if there's no traction, so it reduces power to what it can work with while it tries to get moving and get traction.
