Industrial Vehicle Camera Positioning for Reliable Person Detection

The patent introduces a camera-based visual recognition system as an intermediary to capture environmental images, which are then processed to extract feature points and match with pre-stored reference images. This intermediary approach overcomes the limitations of direct laser radar scanning in environments with similar structures or limited visibility, enabling reliable position determination where traditional methods fail.

The patent replaces the mechanical laser radar scanning system with an optical camera-based system. Instead of using laser beams to measure distances and contours, the system uses cameras to capture images and processes these images computationally to determine vehicle position, thereby adapting to environments where mechanical scanning is ineffective.

The patent creates a digital copy of the environment by capturing reference images with cameras at known positions and storing them with associated position information. This digital map serves as a virtual replica of the physical environment, allowing the system to determine vehicle position by matching current camera images with the stored reference images, eliminating the need for physical position markers.

The patent changes the fundamental parameter for position determination from physical markers (reflective or illuminated objects) to visual features extracted from images (edges, corners, textures). By transforming the position determination problem from a physical marker recognition task to an image feature matching task, the system achieves markerless positioning that works in both internal and external environments.

The patent makes the camera system multi-functional by using the same cameras and image processing pipeline for both vehicle position determination and person detection. The camera captures images that are first used to calculate vehicle position by matching with reference images, and simultaneously used to detect persons by analyzing pixel changes and patterns. This universal approach improves detection reliability without proportionally increasing hardware complexity.

The patent merges the vehicle positioning function and person detection function into a single integrated system. Both functions share the same camera hardware, image processing unit, and computational resources. By combining these functions, the system achieves reliable person detection without the need for separate dedicated sensors, thereby controlling hardware costs and system complexity.