Precise pedicle screw placement is imperative for a range of spinal procedures demanding exact geometric alignment while also carrying inherent risks. The literature reports a complication rate in the case of screw mispositioning of up to 18%. To increase the accuracy of pedicle screw placement, we developed a robotic setup and a breach detection algorithm that could detect a possible perforation of the spinal canal. The robotic setup includes a robotic arm, a drilling system, and specific sensors, e.g., electrical conductivity at the drill bit's tip. The breach detection algorithm consists of a Bayesian-based method providing online and real-time analysis of the electrical conductivity signal to predict a breach. The robotic setup and the perforation detection algorithm were assessed in two ex-vivo experiments. First, data collection was performed by drilling 80 fresh pig vertebrae pedicles, followed by precise data labeling by a surgeon. The evaluation of the proposed algorithm was conducted numerically. Finally, the assessment was performed online by automatically drilling into pedicles in conditions similar to the operating room. The results demonstrated that the algorithm could predict perforations and prevent the robotic setup from going through in 100% of 24 drilled vertebrae. Results demonstrate that using electrical conductivity combined with a robotic setup allows the detection of imminent perforations of the spinal canal during pedicle drilling. This was the first study to evaluate spinal canal perforation detection during ex-vivo pig pedicle drilling.