Robot-assisted laser ablation of bone tissue is an emerging development in orthopedic surgical robotics, but precise control of the ablation depth remains a major challenge for its clinical application. This study achieves feedforward control of laser ablation depth in bone tissue based on pulses-depth relationship models, distinguishing between cortical bone and cancellous bone due to their different ablation mechanisms. Initially, laser ablation experiments were conducted on cortical bone and cancellous bone in vitro, and the experimental data were fitted using a fully connected neural network. Consequently, pulses-depth relationship models for Ho:YAG laser ablation of cortical and cancellous bone were established, specifically mathematical models describing the relationship between the number of laser pulses and the ablation depth. Subsequently, the bone tissue targeted for ablation was classified into cortical and cancellous bone based on bone density. For each category, the corresponding pulse-depth relationship model was applied to determine the required number of laser pulses for the desired ablation depth, thus controlling the initiation and cessation of the laser during the ablation process. To evaluate the accuracy of the pulse-depth relationship models and the effectiveness of the ablation depth control method, laser ablation experiments were conducted following the feedforward control workflow. The results demonstrated that the error between the actual and desired ablation depths was less than 0.8 mm, meeting clinical surgical requirements.