The Navigation of Mobile Robot in the Indoor Dynamic Unknown Environment Based on Decision Tree Algorithm

Area of Science:

• Robotics
• Artificial Intelligence
• Sensor Fusion

Background:

• Mobile robot navigation in dynamic, unknown indoor environments presents significant challenges.
• Inertial Measurement Unit (IMU) yaw values can be prone to errors, impacting localization accuracy.
• Simultaneous Localization and Mapping (SLAM) algorithms require optimization for real-world conditions.

Purpose of the Study:

• To develop and validate an optimized navigation algorithm for mobile robots in indoor, dynamic, and unknown environments.
• To address and correct yaw value errors from IMU sensor fusion.
• To enhance robot pathfinding accuracy and reliability.

Main Methods:

• Analysis of IMU sensor fusion yaw value errors in indoor settings.
• Implementation of adaptive FAST SLAM to optimize odometry-derived yaw values.
• Application of a decision tree algorithm for predicting robot movement direction using fused IMU and SLAM data.
• Design and testing of a physical mobile robot to validate the proposed navigation system.

Main Results:

• The proposed algorithm effectively optimizes yaw values from IMU and odometry data.
• The decision tree-based navigation successfully predicts correct movement directions in complex environments.
• Experimental validation on a real mobile robot confirms the algorithm's effectiveness and validity.

Conclusions:

• The integrated approach of IMU error analysis, adaptive FAST SLAM, and decision tree navigation offers a robust solution for mobile robot navigation.
• The developed algorithm demonstrates high validity and effectiveness in dynamic and unknown indoor environments.
• This research contributes to more reliable and autonomous mobile robot operation.