ModeSleuth

ModeSleuth is a desktop + web app for diagnosing robot dynamics from real log data. You import joint torques, encoder positions/velocities, IMU, and motor currents from ROS bags or CSV, then the app runs automated modal and frequency-response analysis to surface resonances, flexibilities, backlash signatures, and controller-induced oscillations. It outputs actionable reports: suspected mode frequencies per axis, confidence scores, and “what to try next” checklists (filter tweaks, notch candidates, trajectory changes, mechanical inspection hints). This is not a magic auto-tuner; it’s a practical diagnostic tool that shortens the painful loop of “it vibrates, now what?” The UI is built for robotics engineers: time-sync tools, decimation, windowing, coherence plots, and side-by-side comparisons across firmware/controller versions. Export includes plots and parameter suggestions you can paste into common control stacks.