I received some really interesting test data from Lefteris about performance of the mouse sensor.  My basic question was whether a single sensor could track the motion of the robot moving on a curve, and here were his test results -  
 
==========================
Lefteris wrote -  
 
 
As for the sensor I did some tests. I don't know if you remember but the SRV-1 I am working on has a slightly big bias on the left motor. So for me going 99% straight is (53,41). There is unfortunately no 100% value  
 
Going with these values for 1.20 meter gives back values :
dX: +13159  dY: +   48
 
The above means a little bit bias on the left has accumulated in the 1.20 meters that we walked. If it was minus it would mean right. Of course all these are relative to the way you position the sensor.
 
Now let's go like the ones you got in the official firmware. With -15(left),+15(right).
 
With  (53-15,41+15), here are some values as read by the sensor :
 
dX: + 4806  dY: +  627
dX: + 4927  dY: +  634
dX: + 5045  dY: +  646
dX: + 5166  dY: +  659
dX: + 5285  dY: +  671
dX: + 5405  dY: +  685
dX: + 5521  dY: +  701
dX: + 5578  dY: +  708
dX: + 5702  dY: +  722
dX: + 5819  dY: +  737
dX: + 5942  dY: +  754
dX: + 6065  dY: +  768
dX: + 6185  dY: +  784
dX: + 6301  dY: +  794
dX: + 6420  dY: +  805
dX: + 6541  dY: +  815
 
 
You can easily understand that it turned here, towards left.
 
With (53+15,40-15) Here are some values read by the sensor ....
 
dX: + 3028  dY: - 1308
dX: + 3121  dY: - 1334
dX: + 3224  dY: - 1351
 
As you can see ... the turn was REALLY steep the second time because 25 as pwm value for the right motors was ... well low. I think that you can easily understand by the ratio of the forward movement to the Y axis movement and the Y value which way you are going.

Yes that's exactly what I did. Check this picture, even though it's old, almost the same thing applies to my current setup.
 

 
These orange and brown wires go up to the regulator. All I did was just plug a 5V regulator into the battery clip located on the top left of the radio/motor board. As far as the communication between ADNS 2610 and blackfin is concerned I put a zener diode on the DATA line (the other is the CLOCK and it is ran solely by the blackfin) so that whenever the ADNS2610 wanted to send a bit and raised the line to 5V the zener would "switch" it at 3.3V and avoid any potential damage.  
 

Are you sure? I mean in the ADNS2610 datasheet I see that it asks for a single 5V power supply.
 
 
On the other hand it also says this :
 
 
                               MIN                                    MAX
Supply Voltage VDD   -0.5                                    5.5 V
 
Input Voltage          VIN -0.5                           VDD+0.5 V                           SDIO, CLK, LED_CNTL
Input Voltage         VIN -0.5                                 3.6 V                                 OSC_IN, OSC_OUT, REFA
 
Which is quite confusing. To tell you the truth I took it for given that it is 5V since it says supply voltage should be 5V.

Then why not go with that one you just mentioned? The 7050 ? Does anything make 2610 special?  
 
EDIT:
 
Actually I checked the 7050 and it is quite different, it has a lot more pins, and uses a laser diode instead of an LED.

Wow amazing level of integration. It seems like a really nice choice!