DynTuner

DynTuner is a desktop + web app for robotics teams who waste weeks hand-tuning dynamics parameters (mass/inertia, friction, motor constants, delay) to make simulation match reality. You import ROS bag/log data (joint states, IMU, torque/current, commands) and a URDF/SDF model, then run automated parameter identification to fit a dynamics model that predicts measured motion and torques. The app outputs updated parameter sets plus confidence intervals and “what changed” reports so engineers can justify updates. It also generates regression plots and a small validation suite to catch overfitting across trajectories. This is not a magic ‘one-click perfect model’ tool: it focuses on the 20% of parameters that cause 80% of sim-to-real pain, and it flags when your data is insufficient or sensors are too noisy to identify certain parameters.