Cervical arthrodesis surgery focuses on fusing two or more vertebrae using a spinal implant. This paper presents a comanipulation robotic system integrated with an automated deep-learning-based path-planning generator. The cobotic assistance constraints the surgeon’s movements to a predefined drilling direction for precise screw placement. Manually determining the optimal trajectory is challenging and time-consuming due to the need to avoid critical structures such as nerves and vertebral arteries. The proposed approach significantly reduces surgical planning time while improving accuracy and efficiency. Experimental findings indicate promising performance, and the planning outcomes can be applied in surgeries following the surgeon’s assessment. This study highlights the potential of combining robotic assistance and deep learning for safe and effective pedicle screw placement in spinal surgery.