Using the LSTM Neural Network and the UWB Positioning System to Predict the Position of Low and High Speed Moving...

Area of Science:

• Robotics and Autonomous Systems
• Artificial Intelligence
• Sensor Fusion

Background:

• Autonomous transportation relies on accurate positioning systems, but current technologies like Global Positioning System (GPS) and Ultra-Wideband (UWB) have limitations.
• Object displacement during data acquisition and temporarily missing data degrade positioning accuracy and can create safety risks.
• High-speed movement and low position update rates exacerbate the challenge of maintaining real-time, accurate object location awareness.

Purpose of the Study:

• To develop a reliable positioning system for autonomous vehicles by predicting object positions.
• To mitigate the impact of data acquisition delays and sensor limitations on positioning accuracy.
• To enhance safety in autonomous transportation through proactive collision detection.

Main Methods:

• Utilizing a Long Short-Term Memory (LSTM) artificial neural network for position prediction.
• Integrating historical data from Ultra-Wideband (UWB) systems and inertial navigation sensors.
• Developing a predictive model capable of forecasting multiple future positions.

Main Results:

• Achieved a positioning prediction error below 10 cm for both low and high-speed moving objects.
• Successfully predicted up to 10 future positions, providing a significant temporal advantage.
• Demonstrated the system's capability to predict object trajectories and detect potential collisions.

Conclusions:

• The proposed LSTM-based prediction system offers a robust solution for enhancing the reliability of positioning in autonomous transportation.
• Accurate position prediction is essential for ensuring safety and enabling effective collision avoidance in dynamic environments.
• Combining UWB and inertial navigation data with LSTM prediction creates a powerful tool for future autonomous systems.