Using a LRF sensor in the Kalman-filtering-based localization of a mobile robot

Area of Science:

• Robotics
• Sensor Fusion
• Probabilistic Robotics

Background:

• Mobile robot localization is crucial for autonomous navigation.
• The Extended Kalman Filter (EKF) is commonly used for localization with sensors like the Laser Range Finder (LRF).
• Accurate estimation of the output-noise covariance matrix is vital for EKF performance.

Purpose of the Study:

• To propose a novel method for estimating the output-noise covariance matrix in mobile robot localization.
• To compare the computational complexity and statistical accuracy of the proposed method against existing techniques.

Main Methods:

• Utilized the Extended Kalman Filter (EKF) for mobile robot localization using Laser Range Finder (LRF) data.
• Developed a method for estimating line parameter covariances based on classic least squares (LSQ).
• Compared classic LSQ with orthogonal LSQ for computational complexity and statistical accuracy through simulations.

Main Results:

• The classic LSQ method significantly reduces computational complexity compared to orthogonal LSQ.
• The proposed LSQ-based estimation method demonstrates efficiency in localization algorithms.
• Simulations confirm the statistical accuracy of the classic LSQ approach for LRF measurements.

Conclusions:

• The classic least squares (LSQ) method provides an efficient and accurate approach for estimating output-noise covariance in mobile robot localization.
• Implementing LSQ in localization algorithms, particularly within Simultaneous Localization and Mapping (SLAM), leads to reduced computational load.
• This research contributes to improving the performance and efficiency of autonomous mobile robot systems.