Estimation of the angular position of a two-wheeled balancing robot using a real IMU with selected filters

TL;DR

This paper presents a low-cost measurement system for a two-wheeled balancing robot that uses sensors and various filters to accurately estimate the robot's angular position, validated through experimental testing.

Contribution

It introduces a measurement system combining gyroscope, accelerometer, and encoder data with multiple filtering techniques for improved stabilization.

Findings

Kalman filter provided the most accurate angular position estimates.

The measurement system effectively corrected deterministic disturbances.

Experimental validation confirmed the system's reliability and performance.

Abstract

A low-cost measurement system using filtering of measurements for two-wheeled balancing robot stabilisation purposes has been addressed in this paper. In particular, a measurement system based on gyroscope, accelerometer, and encoder has been considered. The measurements have been corrected for deterministic disturbances and then filtered with Kalman, $\alpha$-$\beta$ type, and complementary filters. A quantitative assessment of selected filters has been given. As a result, the complete structure of a measurement system has been obtained. The performance of the proposed measurement system has been validated experimentally by using a dedicated research rig.