Maybe this doesnt exist...
The Wybot M1 Ultra costs $1600 (around twice as much as a comparable corded robot) and is permanently attached by a thick cord to a floating solar-cell/battery power pack. So it doesn't run a cord across your deck, but there's actually more in the pool for swimmers to avoid. And when you remove it from the pool for emptying or to get it out of the way, you need to remove and carry the power pack with it. I have no idea how well it cleans or how reliable it is; I've only seen one review, and it was pretty superficial.
I think a self-contained solar robot would be a lot harder to build:
For a robot that can get into corners and climb walls, there's only about one square foot of available area for a solar panel. In bright sunlight, a solar panel that size would produce around 15 watts. Average peak sunlight duration in the US is 3-5 hours/day; if we're being very generous we could assume the equivalent of 6 hours/day for long summer days. That's 6 x 15 = 90 watt-hours/day.
Battery chargers run at around 80% efficiency. So every day, the robot's battery could be charged to 80% of 90 = 72 watt-hours.
Corded Dolphin robots use 180W; a solar robot with similar capabilities probably would too. So 72 watt-hours would power it for 72/180 = 0.4 hours, or 24 minutes per day.
That sounds almost adequate, but it's a better than best-case estimate. In the real world, 60 watt-hours/day is a more realistic number for clear sunny days, which would only allow 16 minutes of runtime. And on cloudy days, or shorter days with the sun lower in the sky, you might be lucky to get half of that -- so 8 minutes/day of runtime.
And the robot's designers would have to solve some new problems, like how to get it to the surface for charging and then back to the floor for cleaning in a low-power way that doesn't defeat the purpose of having a cordless robot (i.e., without a cord connected to a floating buoy). On the plus side, though, they wouldn't have to worry about how to keep the solar panel clean for maximum efficiency.
Anyway, my overall conclusion from this rough estimation exercise is that current technology would have to improve by about 4x -- putting 4x as much power/day into the battery, or cutting the motors' power requirement by a factor of 4, or some combination of the two -- in order to get a self-contained solar robot that would be adequate for most residential pools.