Detection and Measurement of Object Distances Using the Combination Function of 2D LiDAR Sensor and Camera

Automation systems in transport vehicles such as forklifts can provide solutions in terms of time, quantity, and accommodation costs during operation, as conventional forklift operation requires drivers with specific certifications and qualifications. Automation in forklifts requires various control functions that utilize various types of data. Some functions needed in forklift automation include object detection and object distance measurement. Object distance measurement will utilize LiDAR (Light Detection and Ranging) technology, which offers high accuracy with a wide detection range and can be used at various scanning angles. Using 2D LiDAR for scanning can only provide output in the form of detection along a single horizontal line, so for three-dimensional object recognition, the system requires additional supporting sensors such as cameras. Object detection or recognition using cameras can be performed using deep learning applications with specific datasets of objects to be detected. Object detection is implemented using YOLOv5, while object distance measurement utilizes data fusion and projection from 2D LiDAR with camera data. In this system, object detection data is provided with a precision of 0.97945 using YOLOv5 model with type-m epoch 100, while the best distance measurement is achieved using the projection method from 2D LiDAR data to the camera with an error rate of 0.15%.