For a switching regulator, those requirements are pretty lax--it only has to be 47.5% efficient to pass.
Assuming there are no other requirements (like cost to manufacture), I would just use three separate buck regulators. There are a million ways to implement the switching transistors , but I'd probably do something like the pic:
The top transistor would normally be on all the time, because of R1, but Q2 turns it off by pulling its base down when V2 (a square wave with a 73% duty cycle in this simulation) goes high. When Q1 is on, V1 is "charging" the inductor (you can see the purple trace increasing at this stage). When Q1 goes off, the inductor continues to flow current in the same direction, due to energy stored in its magnetic field, but the current is decreasing because it has to drive the load itself, pulling current through D1.
Note that the top trace is the power dissipated in Q1 (average is about 600mW, which is right at the rating for open-air, but still very safe with one of those little push-on heat sinks). The blue trace is the voltage at the output, while the two bottom traces show the voltage at the collector of Q2 and the emitter of Q1.
You would regulate a circuit like this with pulse width modulation and some kind of feedback signal. The easiest thing to do hardware-wise would probably be to generate your PWM signal with a microcontroller (like a PIC) and feed the output voltage back to one of its analog inputs.