Sensor fusion for robot orientation

Abstract

In dynamic walking environments, robots experience increased body acceleration, making accurate orientation estimation crucial for maintaining balance and control. This work employs an Extended Kalman Filter (EKF) to robustly estimate a walking robot’s orientation by fusing noisy accelerometer and gyroscope data from an Inertial Measurement Unit (IMU). By accounting for system nonlinearities, the EKF significantly enhances real-time attitude tracking and stability, outperforming raw sensor readings and providing a reliable foundation for precise locomotion in complex terrains.