High-precision all-in-one dual robotic arm strategy in oral implant surgery

The study addresses the need for highly accurate dental implant placement—specifically the precise control of implant position, angle, and depth—to ensure long-term success. Traditional static guides lack intraoperative adaptability and can be affected by patient movement, while dynamic navigation systems, although more precise, are costlier and susceptible to occlusions and setup complexity. Robotic assistance offers precision, stability, and efficiency but typically requires extensive calibration and has limited clinical adoption due to high precision demands. The authors propose a high-precision all-in-one, dual-arm robotic system (HADAROIS) to reduce preoperative preparation, mitigate occlusion-induced interruptions, and enable dynamic intraoperative adjustments, thereby improving accuracy, safety, and workflow efficiency in oral implant surgery.

The paper reviews static versus dynamic computer-assisted implant placement. Early dynamic navigation studies (2004–2006; Chiu, Kramer, Brief, Casap) reported angular deviations around 4°. Subsequent research improved accuracy but continued to face occlusion and workflow limitations. Robotic approaches began with Boesecke et al. (MICCAI 2001) demonstrating preoperative 3D planning with robot assistance (1–2 mm positional gaps reported). Sun et al. (2014) described an automated image-guided system with a 6-DOF robot and dental drill, though requiring extensive preoperative CMM correction due to separation from the drilling device. Vision-based navigation for implant robots (Yu et al., 2015) showed promise but suffered from marker occlusion with fixed cameras. The FDA-cleared Yomi system (2017) uses haptic guidance to constrain drill position, direction, and depth, offering high predictability but at high cost and requiring professional oversight. An autonomous implant system (Haidar/Zhao, 2017) was reported but lacks sufficient clinical validation. These works motivate an integrated, occlusion-robust, dynamically navigated robotic solution with reduced setup burden.

System architecture: HADAROIS integrates two robotic arms on a single support body with a control panel. The first arm holds the surgical tool (oral implant handpiece), and the second arm carries a miniature multi-camera image acquisition device for dynamic optical tracking. Universal Robots UR3 manipulators (6 DOF, 3 kg payload, 500 mm reach, ±0.1 mm repeatability) are used; the optical tracker is NDI Polaris Vicra (submillimeter accuracy, ~0.25 mm volumetric accuracy, 95% CI 0.5 mm), integrated close to the surgical field. Coordinate systems: Defined for global (Ow), patient/target (Op), reference component (OA), lens/camera (Ol), second robot (Os), first robot (Oc), and tool (OD), with known transformations to enable precise end-effector localization and path planning. Image acquisition device and OTTM: A custom miniature multi-eye camera (~60 mm diameter, 0.05 mm accuracy, >100 Hz) positioned within ~10 cm of the surgical area minimizes occlusion risk and increases responsiveness. The Occluded Target Tracking Module calibrates internal/external camera parameters, detects multiple markers on patient-worn positioning devices, triangulates 3D marker centers, and declares occlusion when visible markers fall below a threshold. It computes a point-cloud of the scene and searches candidate camera poses within a reachable 3D region by sampling positions and orientations that maintain line-of-sight to all markers; the second arm then repositions the camera to the selected pose to recover tracking. PPDM (implant planning): Using CBCT of the patient wearing a positioning device with dominant spherical markers, the system registers the device to patient anatomy. Along a proposed implant direction L, it samples CBCT values to detect boundary transitions (air–gingiva point a; gingiva–bone points b, c) via gray-value changes. The starting point (b) and an endpoint along L based on the chosen implant length define the drilling trajectory (Fig. 3). PFM (path formulation and cooperative control): Real-time target pose from OTTM is transformed into the first arm’s frame. The PFM determines position, orientation, and speed of both arms, performs collision prediction based on joint and scene geometry, and halts motion if necessary. Clinical workflow (six steps): (1) Acquire CBCT and build 3D model. (2) Determine drilling target with a target-indicating device; install a reference device for head tracking. (3) Recognize marker poses with the binocular camera; register target and reference poses; then track reference markers in real time. (4) Transform target pose to robot frame; generate joint-space trajectory for the first arm. (5) Execute motion: compute joint torques/speeds; perform drilling at the specified pose/path. (6) Closed-loop feedback using joint encoders to match commanded versus actual joint angles. GUI supports device connection, image-mode planning, dental arch selection, and real-time tool visualization.

Simulation experiments: Ten trials on five dental models using a positioning device and planned straight-line drilling yielded an average angular deviation of 1.54° (SD 0.67°) and entry-point deviation of 0.334 mm (SD 0.202 mm), indicating accuracy, consistency, and repeatability. Clinical trials: The system (HADAR/HADAROIS) was used in oral implant surgeries reported for six patients, with detailed results provided for five subjects: angular deviations 0.96°, 0.94°, 1.98°, 2.47°, 1.35°; entry-point deviations 0.26 mm, 0.61 mm, 0.46 mm, 0.09 mm, 0.25 mm. Mean clinical deviations reported were 1.54° (angle) and 0.33 mm (entry point), within clinically acceptable ranges. Visualization of postoperative plan-versus-actual positions shows small angular and entry-point discrepancies. Comparative context: A literature comparison table indicates the study’s mean angular and entry-point deviations are lower than several published reports (e.g., Ersoy 4.25°, 0.99 mm; Cristache 2.46°, 0.79 mm; Chmielewski 2.73°, 0.48 mm), suggesting favorable accuracy for the proposed system. Operational advantages: The integrated dual-arm design reduced preoperative calibration needs, allowed occlusion-robust tracking via OTTM, and enabled cooperative motion with PFM for collision avoidance, improving workflow efficiency and safety.

The proposed integrated dual-arm HADAROIS system directly addresses the challenges of lengthy preoperative calibration, static planning constraints, and intraoperative occlusion in dynamic navigation. By co-locating the surgical tool and a maneuverable, close-range multi-camera tracker on coordinated arms, the system maintains continuous, precise tracking and enables real-time plan adherence and adjustments. The PPDM leverages CBCT-derived tissue boundaries to streamline and standardize planning, reducing dependence on operator experience. The PFM ensures coordinated, collision-free motion, enhancing safety. Empirical results from simulation and initial clinical use demonstrate angular and entry-point deviations within accepted clinical thresholds and compare favorably to prior studies, supporting the system’s efficacy in accurate implant placement. These findings suggest that an all-in-one, occlusion-robust robotic approach can increase precision and efficiency in implant dentistry while improving intraoperative robustness.

The study presents an integrated, dual-arm high-precision navigation and positioning system for oral implant surgery, accompanied by cooperative control strategies (OTTM, PPDM, PFM). This all-in-one design shortens preoperative preparation, supports dynamic intraoperative adjustments, and provides accurate guidance for drilling trajectories. Simulation and clinical results demonstrate low angular and entry-point deviations, indicating reliable, precise implant placement. The system offers a promising, efficient, and safe solution for improving outcomes in dental implant procedures, with potential for broader clinical adoption and further performance refinements.

Explicit limitations are not extensively discussed. The clinical evaluation reports detailed outcomes for five subjects (with mention of six surgeries), representing a small cohort without long-term follow-up or a randomized comparator, which may limit generalizability. Some reporting inconsistencies (e.g., averages in different sections) are noted. Future work is planned to further improve performance and address issues encountered during surgery.