State Transition for Statistical SLAM Using Planar Features in 3D Point Clouds

Area of Science:

• Robotics
• Computer Vision
• Artificial Intelligence

Background:

• Simultaneous Localization and Mapping (SLAM) is crucial for autonomous vehicles.
• Current SLAM solutions often rely on statistical filters like the Kalman filter, using simple point features.
• Advancements in 3D scanning provide richer data for more sophisticated SLAM approaches.

Purpose of the Study:

• To propose a novel approach for SLAM using planar features within multi-object Bayesian filters.
• To develop a stochastic transition model for state prediction in Bayesian filters.
• To evaluate the accuracy and efficiency of the proposed method for autonomous vehicle applications.

Main Methods:

• Development of a stochastic transition model for Bayesian filters.
• Implementation of a state prediction solution utilizing planar features.
• Simulation studies using real vehicle sensor data.

Main Results:

• The proposed model successfully predicts future planar features and vehicle states.
• Demonstrated reasonable accuracy in state prediction for SLAM.
• Showcased efficiency for statistical filtering-based SLAM applications.

Conclusions:

• Planar features can be effectively utilized in multi-object Bayesian filters for SLAM.
• The developed state prediction model shows promise for enhancing autonomous vehicle navigation.
• The approach offers a viable and efficient alternative for statistical filtering-based SLAM.