Camera and LiDAR Sensor Fusion Method for Object Estimation of Intelligent Robots

This study proposes a low-cost camera and LiDAR sensor fusion method to improve object recognition and estimation performance in autonomous and robot systems. We first calibrate the camera and LiDAR sensor to implement the proposed method. Then, data association is performed between the bounding box data of the object identified through the YOLOv3 algorithm from the camera sensor and the LiDAR sensor data. Afterward, a Kalman filter considering sensor characteristics is applied to improve object recognition and estimation accuracy. The usefulness of the proposed method was validated through scenario-based experiments. Experimental results revealed that the proposed sensor fusion method improved object recognition and estimation accuracy, compared to methods using single sensors.