Pedicle screw placement (PSP) is a common procedure for spine surgery with proven benefits. However, some drawbacks exist; the screw can be misplaced or may loosen. Inaccurate screw placement may cause neurological impairment such as pain, weakness, or sensory loss. Thus, screw placement accuracy is of utmost importance and is critical for a successful surgery [1]. Computer-guided systems, such as fluoroscopy-based navigation, are primarily applied to the lumbar spine. In recent years, ultrasound (US) navigation systems have been developed and investigated for use in spine surgery. This eliminated the need for multiple fluoroscopic images to update instrument positioning and significantly improved pedicle screw insertion. However, intraoperative navigation requires human to be in the loop and manual manipulation. Therefore, adopting a robot-assisted system is necessary for pedicle screw placement. Several studies have been proposed to assist in pedicle screw placement with a robot-assisted system. Smith et al. also implemented a camera-based supervisory controlled autonomous robot for polyaxial spine screw placement [2]. They reported an entry point accuracy of 0.49 ± 0.17 mm and a destination point accuracy of 1.49±0.46 mm compared to the CT model registered with the attached reference marker. However, the optical system suffers from line-of-sight problems. This paper aims to develop a non-radiation robot-assisted system for pedicle screw placement. The preoperative surgical plan is registered in the 3D US reconstructed model to achieve this. The robot could then automatically conduct the drilling procedure by replacing the US probe. Finally, this system is validated with an ex-vivo lamb spine. To the best of our knowledge, this is the first work to perform robot-assisted PSP without an external tracking system.