Effects of Systemic Error for Localization and Control of Differential Drive Mobile Robot

---

Abstract

A novel exploration into the practical application of the Kalman filtering technique for the control and observability of a differential drive mobile robot is presented in this article. The effectiveness of Kalman filtering in addressing the localization challenges within a partially known environment is assessed. In a controlled quasi-static setting, odometric error data is meticulously collected on a metric scale, and deviations are analyzed to establish a pseudo-random error model for a differential drive robot under two scenarios: a fully functional system and a system experiencing partial failure. Surprisingly, it is found that despite calibrated odometric readings, the controller struggles to differentiate between systemic and non-systemic errors, treating both as identical. This unique challenge manifests when one or more sensors/actuators introduce non-systemic errors perceived as systemic errors. The study extends to mobile robots equipped with ultrasonic sensors, precisely delineating ranges within ±2 cm along the heading direction, providing valuable insights into the nuanced dynamics of Kalman filtering and paving the way for future advancements in mobile robot control systems.

---

Keywords

Autonomous Mobile Robot (AMR); Localization; Kalman Filter; Odometry; Systemic Errors Sensor Fusion;