The moon running robocar has taken it's first tour according to a simple preset path of right turn, left turn, forward and backward motion of one second duration each. It was a four second long trip. The turn motions were complete disasters, mainly due to the way it is designed to turn, but the actual assembly won't allow it. Can you imagine a car turning on its place, only using two rear wheels spinning at opposite directions? That's exactly what happens with the current assembly and it's going to be fixed soon. Other than that, the robocar is running pretty good. You can watch the video by the way.
Regardless, the first battery powered test drive was a success. It gave valuable feedback on what should be done next. The design of the drive train compartment will be updated to accommodate four motors instead of two. Of course a two motor caterpillar track drive is an option and probably will look better too, but... I decided to use only 3D printed parts. That includes wheels. Ready made wheels are cheap and probably I will uses those later, but until the thing is finalized, I don't think that is ok. I mean, I want to be flexible about the drive train. There are several variations:
1) four fixed wheels, four motors, tank like motion
2) multiple wheel tank caterpillar track style
3) four wheel caterpillar track style
4) two wheel drive, front steering, traditional car style
Of course, more options are possible, but the point here is the accuracy of motion rather than any other factor.
Accuracy and Cost
It's not that accuracy can be sacrificed to save at cost. Does not work that way. It's about maintaining accuracy while spending less. Tank style is the most accurate but costly. Four wheel/four motor is the cheapest.
Accuracy and Complexity
That is probably the biggest challenge I will have to face along the way. Accuracy is impossible without data analysis. Data analysis requires input. Input is impossible without the proper hardware. That brings us down to using some sort of sensors. Extra sensors require extra IO pins. Because I am using Arduino Nano, I am already using all pins, but today I learned about IO pin expansion boards and I am going to use one to expand. I may be able to use a TFT screen I have felt the urgency of during testing. User needs a visual feedback more meaningful than a few LEDs and an LCD screen is exactly that.
Next comes the source code upgrade to accommodate the expansion board related changes.