CASE STUDY
ariadne · ESP32 micromouse
Control and planning for an ESP32 micro-robot: 2.4 times faster than stop-and-pivot in simulation.
SPEED-UP / STOP-PIVOT×ariadne · in simulation
CONTINUOUS INTEGRATIONtestsariadne · C++ and Python
01 · Problem#
Get a small robot through a maze as fast as possible, without slipping or losing its path.
02 · Approach#
- Clothoid turns and time-optimal speed planning.
- Pole-placement control, validated on a physical model simulated at 10 kHz.
- Embedded C++ planner (FreeRTOS) bit-for-bit identical to its Python reference.
03 · Result#
- 2.4 times faster than the stop-and-pivot strategy, in simulation.
- 115 tests in continuous integration.
04 · What it shows#
Control and real-time embedded work: model, tune a controller, then prove with tests that the embedded code does exactly what the model predicts.
05 · Limitations#
Results obtained in simulation.