SENSOR

• We assumed that the IR reflective sensors were linear, but this was not the case. Linearization proved possible (and was done), but hard.
• We corrected the non-linearity of the SHARP sensors with the values from the datasheet, but the sensor’s real readings differed substantially from those mentioned in the datasheet. This was possibly due to surrounding influences.

CIRCUITS

It is absolutely necessary to check every circuit first on a breadboard. We didn’t always do this which resulted in some mistakes before re-testing.

• First, we made our buffers wrong. Some time had passed before we realized it was not the opamps that were malfunctioning, but the circuit itself had been designed wrong.
• For the IR reflective sensors: we first amplified the signal and then subtracted the offset, however, the opamps saturated every time. We eventually solved this problem by first subtracting the offset before amplifying the signal.
• In order to make -12V, we used a circuit design we had found on the Internet. However, the used microchip broke down about two days before the final deadline. A back-up chip was available, but we decided against using it before we had figured out why the first one had broken down. This took us a few hours: it turned out we had to place very fast capacitors to protect the transistor in the chip.

PROGRAM

• In our opinion, the execution speed of the programme, which was deemed so important that the programme should be written entirely in Assembler, was much less critical than assumed by our professor. The microcontroller executed one instruction every microsecond. It might have been possible to write the programme in a high(er) level language. An added advantage would have been that, since writing the programme itself would have been easier (more understandable syntaxis and programme structuring), it might have been possible to put more effort into the contents of the programme itself.
  Given more time, we might have tried re-writing the programme in (e.g.) C and then observe whether there was a difference in swiftness of reaction of the robot. Due to lack of time, this was unfortunately not possible. However, if the soccer playing robot project is further developed in the future, it should be re-considered whether re-writing the program in a higher level language could be worth the effort.