Small Imaging Depth LIDAR and DCNN-Based Localization for Automated Guided Vehicle

Area of Science:

• Robotics and Automation
• Sensor Technology
• Computer Vision

Background:

• Automated Guided Vehicles (AGVs) require precise localization for efficient operation.
• Small imaging Light Detection and Ranging (LIDAR) and fusion-based localization are critical for AGV systems.
• Existing systems often face challenges with calibration, especially in dynamic environments.

Purpose of the Study:

• To introduce a novel, calibration-less imaging LIDAR sensor and a fusion-based localization method for AGVs.
• To enhance the localization accuracy and target detection capabilities of AGVs.
• To develop a robust system suitable for both indoor and outdoor AGV applications.

Main Methods:

• Development of a Single-Photon Avalanche Diode (SPAD) LIDAR sensor utilizing time-of-flight and SPAD arrays.
• Integration of dual SPAD arrays for simultaneous range and monocular image data acquisition in a unified coordinate system.
• Implementation of a Deep Convolutional Neural Network (DCNN) based fusion method (SPAD DCNN) to process LIDAR outputs.

Main Results:

• The SPAD LIDAR sensor provides simultaneous, calibrated range and monocular image data.
• The SPAD DCNN method effectively fuses range, monocular, and peak intensity image data.
• Experimental evaluation demonstrated improved localization accuracy for AGV trajectories.

Conclusions:

• The developed SPAD LIDAR sensor and SPAD DCNN localization method offer a significant advancement for AGV systems.
• The calibration-less design and fusion capabilities enhance robustness and accuracy in challenging AGV navigation scenarios.
• This technology holds promise for improving the performance and reliability of AGVs in various industrial and logistical applications.