inertial-navigation

C++ integrated navigation library. Inspired by Principles of GNSS, Inertial, and Multisensor Integrated Navigation Systems, Groves, P.

Header-only C++17/Eigen strapdown INS library: Initial alignment, Rodrigues DCM mechanization, Coriolis-compensated velocity integration, and Van Loan error covariance propagation.

GPS-denied pedestrian dead reckoning: learned inertial velocity fused in a 2D error-state Kalman filter. SIH 2026 · SIH26168 · ISRO.

GNSS-denied vehicle navigation: strapdown INS, 15-state error-state Kalman filter, and a drift benchmark. 5.36% to 0.22% drift over a 120s outage.

Multisensor navigation and georeferencing using GNSS, INS and photogrammetry for trajectory estimation.

Smartphone dead reckoning that survives GNSS loss. A 240 KB CNN reads vehicle speed out of phone IMU vibration and feeds a Kalman filter with its own uncertainty: 1.5% median drift over 600-850 m outages, 30/30 held-out rides under 10%. Android app, Python engine, and all 71 minutes of ride data. SIH 2026, Team OFFGRID.

NASA Space Apps Challenge 2016 — Regional Champion (Dhaka), Global Nominee. Inertial navigation device that tracks movement from a start point and backtracks the path when GPS and comms are unavailable.

Open-source temperature-controlled MSPM0 + ICM-45686 IMU with CAN/UART and Windows Studio

Reconstructing a walked path from a phone IMU alone, and measuring the inertial drift: 5.14 m over a 76.2 m loop-closed route

Python library for IMU processing: Initial alignment (gravity leveling, gyrocompassing), continuous attitude tracking via Rodrigues rotation, and DCM/Euler angle kinematics.

Why 84.4 minutes makes a pendulum immune to acceleration. An interactive, first-principles simulation of Schuler tuning and the error dynamics that make inertial navigation possible.

Header-only C++17/Eigen 15-state EKF for tightly-coupled UWB/IMU indoor positioning: ZUPT, magnetometer heading, height constraints, Joseph-form covariance updates.