[ros-users] [Discourse.ros.org] [TurtleBot] Absolute motion in serial protocol

[Daniel Stonier]

Could, should, would have ... 

The program inside the microcontroller is essentially the same, with minor modifications, to that used in Yujin's cleaning robot products. There are a few reasons why it's nonsensically expensive to make too many changes to microcontroller code within a company:

1) RFC (Request for Change) processes need to be followed to any modification of the flashed program at the factory. These processes are typically long, require a significant round of testing and QA to approve.
2) If a bug gets through the cracks, you inflict re-flashing on your company's partners and users and then you have to deal with a very very (did I already say very?) unhappy CTO/Business Director. Alternatively, inflicting a deb update on users is something that people rarely even notice has happened.
3) Usually there's only one person with the knowledge (more would be a waste) to develop the microcontroller code with its esoteric toolset AND that person is on the product team. You should be thankful for the few brief stints he has available to contribute to a research robot!
4) Expanding the serial protocol is more work than just creating a ROS message and developers, unless motivated by the appropriate carrots, will always choose the path of least resistance.

We would have loved to make that code open source so others could have evolved the controller. Unfortunately given the history of that codebase, this was never going to be possible without completely rewriting it.





---
[Visit Topic](https://discourse.ros.org/t/absolute-motion-in-serial-protocol/4417/2) or reply to this email to respond.