Robotic technologies have promising applications in computed tomography (CT)-guided puncture. However, nodule-surrogate models can be difficult to develop for relevant studies, and thus the accuracy of optical navigation robot-assisted puncture remains unclear. This study aims to evaluate a starch mixture (the starch group) and a copper particle nodule-surrogate model (the particle group) and to compare the accuracy of optical navigation robot-assisted puncture (the robot group) with traditional CT-guided manual puncture (the manual group) using swine liver and kidneys.The study was approved by the institutional animal care and use committee. Ex vivo and in vivo studies of three swine liver and kidney samples using nodule surrogates were imaged by CT scan to assess the accuracy of the starch and particle groups. In an in vivo study of six swine, 24 punctures made by the robot and manual groups were performed using copper particle nodule-surrogate targets in the liver and kidneys under CT guidance. The accuracy of insertion was evaluated with a 5.0-mm margin. The needle insertion time, level of radiation exposure, and complications were evaluated.In the first experiment, all nodule surrogates were easily visible on the CT images. However, other aspects of the starch group (one starch overflow, one starch dispersion event, and one air embolism) were inferior to those of the particle group. In experiment 2, the accuracy of needle insertion in the robot group (3.71±1.34 mm) was higher than in the manual group (11.89±9.59 mm) (P<0.001). The needle insertion time and level of radiation exposure were superior in the robot group compared to the manual group. Complications were similar between the two groups.The particle method may be superior to the starch method. The robot group was more accurate than the manual group, and the occurrence of complications was similar.Copyright © 2025 AME Publishing Company. All rights reserved.