I can "solve" a lot of traffic problems too, if nobody involved is worried about collisions.
Though based on my car's adaptive cruise control ability, we've got a long way to go before we can have cars that drive within a few feet of each other at highway speeds, and even longer before we could have intersections without signals (or stops).
All it takes is one rock dropping onto the road heavy enough to immediately stop a car in its tracks, and what would follow would be a series of collisions affecting a large number of cars.
Then you need to leave gaps for new arrivals from a different road, and in the end I don't think the efficiency would be that much higher.
I think the benefits are more that the car strictly obeys road rules, as opposed to humans being unaware, ignoring, or straight up willfully breaking (e.g. speeding) the rules.
<rant>
On my path to and from work there is a "trouble nexus" - you have a few main roads intersecting, a relatively short stretch of road with a couple of entrance/exits towards a public transit hub (trains, buses, and a large carpark converging - so there's a lot of road traffic), then a bridge, then another major intersection and exits onto a highway.
People are always slowing down to create a gap in the traffic flow and let vehicles from the transit hub onto the road leading up to the bridge, in an effort to be kind towards other drivers, but this has disastrous effects on the traffic behind it - interrupting the traffic flow of the major roads.
If people did not make space for vehicles from the side roads, the main flow of traffic would pass fairly quickly and then the sideroad traffic would be able to get on noramlly.
</rant>
The point I am trying to make is that autonamous cars would let traffic planners adjust e.g. road speeds, traffic light timings, etc. closer to a global optimum without having to account for human selfishness.
>"Traffic jams are ubiquitous in human society where individuals are pursuing their own personal objective," the authors explain. "In contrast, ants share a common goal: the survival of the colony, thus they are expected to act cooperatively to optimise food return."
If everyone drove 40mph during rush hour, we'd be well within capacity even with increased drivership. When everyone tries to go 65+mph, that's when things slow down to 5mph. Prisoner's dilemma here, not the limits of acceleration.
Momentum v strength doesn't scale linearly.
Just some bad highway design. Also places like the Embarcadero interchange at the 101 in Palo Alto has existing traffic fighting with entering traffic within 200 meters of space. HOV lanes make it even worse because you end up squeezing all the lanes and you create incentives for HOV drivers to merge all the way to the left as quickly as possible which further delays the main lanes since people are getting slowed down from the lane changes. Then, when HOV drivers have to exit, again they have to merge all the way right again. There is a bottleneck squeeze and the lateral squeeze happening as well. Houston traffic by comparison is “bad” but far better than the Silicon Valley version despite Silicon Valley having a smaller population and fewer cars.
Exit before/ merge after we much better than exit after/merge before. SV has many of these congestion building ramp strategies.
So an ant can go from full speed, to stopped, to full speed in a fraction of a second. A car on the highway will take several seconds to stop, and much longer to smoothly accelerate, leading to standing wave type effects on highways. This is much less likely with people walking, and even less likely with ants.
The dynamics of cars on the highway is governed by the difficulty of changing speed and direction. For ants, this constraint is negligible. So these things don't really compare.