IMU Calibration & Self-Correction Algorithm for Open Source Autonomous Vehicle Controller

Information obtained from an Inertial Measurement Unit (IMU) is constrained due to physical limitations such as noise, drifts, misalignments and offsets. Since obtaining the state of a robot/vehicle requires integrating IMU outputs with time, the errors in the integrated estimates grow very fast. This project develops tools to accurately model the errors that affect IMU outputs, to estimate their relevant parameters and to understand how each parameter changes with time. Additionally, a sensor-fusion based pipeline is proposed for regularly correcting the estimated IMU parameters. The proposed method does not require any external mechanical setup and is designed to ideally work on low-computation processors and/or microcontrollers.