dotbot/dotbot_simulator: add logic to load custom C based control loop function

[aabadie]

In simulator_init_state.toml, just add to each robot configuration the option custom_control_loop_library to specify the custom control loop library to use. Example:

[[dotbots]]
address = "BADCAFE111111111" # DotBot unique address
calibrated = 0xff # optional, defaults to only first lighthouse calibrated
pos_x = 400 # [0, 2_000] in mm
pos_y = 400 # [0, 2_000]
theta = 0.0 # [0.0, 2pi]
custom_control_loop_library = "librobotcontrol.so"  # <= line to add

The control loop is set per robot so each robot in the swarm can use its own control loop library. The default is to use the Python based version already available in the code base.

[codecov]

Codecov Report

❌ Patch coverage is 36.36364% with 56 lines in your changes missing coverage. Please review.
✅ Project coverage is 68.79%. Comparing base (a22c58e) to head (b0ee900).
⚠️ Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
dotbot/dotbot_simulator.py	36.36%	56 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #222      +/-   ##
==========================================
- Coverage   69.25%   68.79%   -0.47%     
==========================================
  Files          40       40              
  Lines        3789     3829      +40     
==========================================
+ Hits         2624     2634      +10     
- Misses       1165     1195      +30     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.

[geonnave]

Very interesting!

Just confirming I understood: with this we can copy/paste the update_control library to/from the dotbot firmware code and compile it with target=host, then use it from the simulator, is that it?

If so, would it make sense to also provide an example C file (and Makefile) in the examples folder?

[aabadie]

with this we can copy/paste the update_control library to/from the dotbot firmware code and compile it with target=host, then use it from the simulator, is that it?

The idea is to compile the source that is located in dotbot-firmware (or libs). There's no code to copy. The output library (.so or dylib) can be anywhere. The path to the .so/dylib is set in the simulator configuration file. I indeed plan to provide a helper Makefile or CMake project to compile the library and I'll probably move the .h/.c code in a segger library under DotBot-libs so it can be reused in dotbot-swarmit as well.

[aabadie]

@geonnave I added a513149 which contains a CMake project to build the library locally. The readme explains how to use it.

[WilliamTakeshi]

Shouldn't this check be before the

        self.custom_robot_control.waypoint_x = int(
            self.waypoints[self.custom_robot_control.waypoint_idx].pos_x
        )
        self.custom_robot_control.waypoint_y = int(
            self.waypoints[self.custom_robot_control.waypoint_idx].pos_y
        )

a couple lines earlier? It will break a couple lines back if self.custom_robot_control.waypoint_idx >= len(self.waypoints)

[WilliamTakeshi]

This is generating a .TOML right? but latter the toml read looks for this field:

custom_control_loop_library: Path = None. it doesn't match

[WilliamTakeshi]

I didn't knew what this does, so I asked chatGPT and he said this:

Without argtypes/restype, ctypes assumes restype=c_int and performs no argument validation. On x86-64 Linux this often "works" by accident, but on other platforms (macOS ARM, Windows) or
with different calling conventions, wrong assumptions corrupt the stack. The C function update_control almost certainly returns void.
Fix: After loading the CDLL, declare the signature:
self.custom_control_loop_library.update_control.argtypes = [ctypes.POINTER(RobotControl)]
self.custom_control_loop_library.update_control.restype = None

[WilliamTakeshi]

I dont know how it is in python, but in rust we have to use #[repr(C)] . Do we have to align with the C code?