Real-time active constraint generation and enforcement for surgical tools using 3D detection and localisation...

Area of Science:

• Robotics
• Surgical Technology
• Computer Vision

Background:

• Collaborative robots (cobots) benefit from active constraints for safety during human-robot interaction.
• Current systems face a gap between bulky haptic feedback robots and miniaturized systems with limited constraint enforcement.
• Active constraint generation often relies on external tracking and preoperative imaging.

Purpose of the Study:

• To present a refined collaborative robot platform for orthopaedic surgery.
• To introduce an 'on-the-fly' active constraint generation and enforcement method using a monocular RGB camera network (SimPS-Net).
• To evaluate the system's effectiveness in defining and enforcing safe and restricted regions.

Main Methods:

• A refined three-degrees-of-freedom collaborative robot, the Signature Robot, was utilized.
• A real-time monocular RGB camera-based network (SimPS-Net) was employed for on-the-fly boundary definition.
• Two active constraints (safe region, restricted region) were tested for enforcement.

Main Results:

• The SimPS-Net achieved a 54.7% ± 5.2% object localization success rate.
• In the safe region, haptic feedback resisted boundary تجاوزs (2.70 mm ± 0.37 mm from boundary).
• The restricted-zone constraint successfully prevented boundary penetration in 100% of attempts.

Conclusions:

• The proposed robotic platform and constraint pipeline demonstrate viability for surgical applications.
• Real-time, camera-based active constraint generation and enforcement are effective.
• The system shows promise for enhancing safety in collaborative surgical robotics.