So why are photoelectric panels flat? In this post Why not use our own light production to produce new energy instead of wasting it?, I naively asked if it was possible to also recycle our own lightning at night.
someone, in his answer, also noticed that photoelectric cells are highly angle dependent.
Hence my first thought: why the photoelectric cells are not designed to be hemispheric?
EDIT1: people misunderstood, so I'm correcting: I mean the panel would be flat, parallel to the ground for example, but each photoelectric cell would be hemispheric. Thus getting optimum direct radiance over time from the sun as also irradiance from the whole sky. As side-effect, light from every point and also because cells are highly reflective, including sunlight would also bounce multiple times from one cell to the others around before dissipating instead of being reflected toward the sky. So does it worth it?
 A: The reason is simple. If you make it hemispherical, and shine sunlight vertically on it, less light will reach the photoelectric cells ( due to the formation of shadows )! So, if we put it horizontally, there will be no shadows and most of the light will be used for the desired purpose and in the other case, much of the light is wasted!
A: The photocell itself is far easier to produce when flat, although bendable sheets (allegedly for clothing!) are in development.
However, as the other answers point out,  curving the photocell doesn't help.  The most common way of maintaining the cell's perpendicularity to the sun is with external servos to rotate the array.   There is an alternative, used in some digital cameras:  mount a small hemispherical lenslet over each element (solar cell or camera pixel) to partially focus light onto the sensitive area.  For cameras, this has the added advantage that photons reflected off the pixel are likely to suffer total internal reflection at the lenslet's surface and then get a "second chance" to be absorbed at the pixel.  
For solar cells, I suspect the difficulty in maintainence of such lenslets overwhelms and possible collection advantage.
A: I've been studying solar panels so I hope I can clarify your doubts.
The amount of solar irradiance which actually interest your surface (the panel) is dependent on the angle between your surface and the sun beams. This means that it is both affected by the titl angle of the panel and by the sun daily and seasonal position.
Solar panels angles are then usually optimized (in terms of their tilt angle) in order to absorb the maximum amount of solar irradiance over the year.
As CuriousOne tried to explain, if you have emispheric cells it means that you don't have a titl angle for the panel, but you rather have infinite tilt angles for any infinitesimal point of the emispheric surface! Only one point of the surface will get the maximum irradiance, while others will be penalised, even if neglecting shadowing effects and other stuff. I've understood that you are talking about a FLAT panel made of many emispheric cells, but you would have the problem I've explained for any single emispherical cell on the panel.
A: Easier to manufacturer.  To add a curve to panel to capture the sun's photons better would very fractional and not be cost efficient to build vs added energy.  
