Implantology is widely considered the preferred treatment for patients with partial or complete edentulous arches.34,35 The success of the surgery in achieving good esthetic and functional outcomes is directly related to correct and prosthetically-driven implant placement.36 Accurate implant placement is crucial to avoid potential complications such as excessive lateral forces, prosthetic misalignment, food impaction, secondary bone resorption, and peri-implantitis.37 Any deviation during the implant placement can result in damage to the surrounding blood vessels, nerves, and adjacent tooth roots and even cause sinus perforation.38 Therefore, preoperative planning must be implemented intraoperatively with utmost precision to ensure quality and minimize intraoperative and postoperative side effects.39

Currently, implant treatment approaches are as follows: Free-handed implant placement, Static computer-aided implant placement, and dynamic computer-aided implant placement. The widely used free-handed implant placement provides less predictable accuracy and depends on the surgeon’s experience and expertise.40 Deviation in implant placement is relatively large among surgeons with different levels of experience. When novice surgeons face complex cases, achieving satisfactory results can be challenging. A systematic review41 based on six clinical studies indicated that the ranges of deviation of the platform, apex, and angle from the planned position with free-handed implant placement were (1.25 ± 0.62) mm–(2.77 ± 1.54) mm, (2.10 ± 1.00) mm–(2.91 ± 1.52) mm, and 6.90°± 4.40°–9.92°± 6.01°, respectively. Static guides could only provide accurate guidance for the initial implantation position. However, it is difficult to precisely control the depth and angle of osteotomies.42 The lack of real-time feedback on drill positioning during surgery can limit the clinician’s ability to obtain necessary information.42,43,44 Besides, surgical guides may also inhibit the cooling of the drills used for implant bed preparation, which may result in necrosis of the overheated bone. Moreover, the use of static guides is limited in patients with limited accessibility, especially for those with implants placed in the posterior area. Additionally, the use of guides cannot flexibly adjust the implant plan intraoperatively. With dynamic computer-aided implant placement, the positions of the patient and drills could be tracked in real-time and displayed on a computer screen along with the surgical plan, thus allowing the surgeon to adjust the drilling path if necessary. However, the surgeons may deviate from the plan or prepare beyond it without physical constraints. During surgery, the surgeon may focus more on the screen for visual information rather than the surgical site, which can lead to reduced tactile feedback.45 The results of a meta-analysis showed that the platform deviation, apex deviation, and angular deviation were 0.91 mm (95% CI 0.79–1.03 mm), 1.26 mm (95% CI 1.14–1.38 mm), and 3.25° (95% CI 2.84°–3.66°) respectively with the static computer-aided implant placement, and 1.28 mm (95% CI 0.87–1.69 mm), 1.68 mm (95% CI 1.45–1.90 mm), and 3.79° (95% CI 1.87–5.70°), respectively, with dynamic computer-aided implant placement. The analysis results showed that both methods improved the accuracy compared to free-handed implant placement, but they still did not achieve ideal accuracy.46 Gwangho et al.47 believe that the key point of a surgical operation is still manually completed by surgeons, regardless of static guide or dynamic navigation, and the human factors (such as hand tremble, fatigue, and unskilled operation techniques) also affect the accuracy of implant placement.

Robotic-assisted implant surgery could provide accurate implant placement and help the surgeon control handpieces to avoid dangerous tool excursions during surgery.48 Furthermore, compared to manual calibration, registration, and surgery execution, automatic calibration, registration, and drilling using the dental implant robotic system reduces human error factors. This, in turn, helps avoid deviations caused by surgeons’ factors, thereby enhancing surgical accuracy, safety, success rates, and efficiency while also reducing patient trauma.7 With the continuous improvement of technology and reduction of costs, implant robotics are gradually becoming available for commercial use. Yomi (Neocis Inc., USA) has been approved by the Food and Drug Administration, while Yakebot (Yakebot Technology Co., Ltd., Beijing, China), Remebot (Baihui Weikang Technology Co., Ltd, Beijing, China), Cobot (Langyue dental surgery robot, Shecheng Co. Ltd., Shanghai, China), Theta (Hangzhou Jianjia robot Co., Ltd., Hangzhou, China), and Dcarer (Dcarer Medical Technology Co., Ltd, Suzhou, China) have been approved by the NMPA. Dencore (Lancet Robotics Co., Ltd., Hangzhou, China) is in the clinical trial stage in China.

Basic research on dental implant robotic system

Compared to other surgeries performed with general anesthesia, dental implant surgery can be completed under local anesthesia, with patients awake but unable to remain completely still throughout the entire procedure. Therefore, research related to dental implant robotic system, as one of the cutting-edge technologies, mainly focuses on acquiring intraoperative feedback information (including tactile and visual information), different surgical methods (automatic drilling and manual drilling), patient position following, and the simulation of surgeons’ tactile sensation.

Architecture of dental implant robotic system

The architecture of dental implant robotics primarily comprises the hardware utilized for surgical data acquisition and surgical execution (Fig. 4). Data acquisition involves perceiving, identifying, and understanding the surroundings and the information required for task execution through the encoders, tactile sensors, force sensors, and vision systems. Real-time information obtained also includes the robot’s surrounding environment, object positions, shapes, sizes, surface features, and other relevant information. The perception system assists the robot in comprehending its working environment and facilitates corresponding decision-making as well as actions.

Fig. 4

The architecture of dental implant robotics

During the initial stage of research on implant robotics, owing to the lack of sensory systems, fiducial markers and corresponding algorithms were used to calculate the transformation relationship between the robot’s and the model’s coordinate system. The robot was able to determine the actual position through coordinate conversions. Dutreuil et al.49 proposed a new method for creating static guides on casts using robots based on the determined implant position. Subsequently, Boesecke et al.50 developed a surgical planning method using linear interpolation between start and end points, as well as intermediate points. The surgeon performed the osteotomies by holding the handpieces, with the robot guidance based on preoperatively determined implant position. Sun et al.51 and McKenzie et al.52 registered cone-beam computed tomography (CBCT) images, the robot’s coordinate system, and the patient’s position using a coordinate measuring machine, which facilitated the transformation of preoperative implant planning into intraoperative actions.

Neocis has developed a dental implant robot system called Yomi (Neocis Inc.)53 based on haptic perception and connects a mechanical joint measurement arm to the patient’s teeth to track their position. The joint encoder provides information on the drill position, while the haptic feedback of handpieces maneuvered by the surgeon constrains the direction and depth of implant placement.

Optical positioning is a commonly used localization method that offers high precision, a wide -field -of -view, and resistance to interference.54 This makes it capable of providing accurate surgical guidance for robotics. Yu et al.55 combined image-guided technology with robotic systems. They used a binocular camera to capture two images of the same target, extract pixel positions, and employ triangulation to obtain three-dimensional coordinates. This enabled perception of the relative positional relationship between the end-effector and the surrounding environment. Yeotikar et al.56 suggested mounting a camera on the end-effector of the robotic arm, positioned as close to the drill as possible. By aligning the camera’s center with the drill’s line of sight at a specific height on the lower jaw surface, the camera’s center accurately aligns with the drill’s position in a two-dimensional space at a fixed height from the lower jaw. This alignment guides the robotic arm in drilling through specific anatomical landmarks in the oral cavity. Yan et al.57 proposed that the use of “eye-in-hand” optical navigation systems during surgery may introduce errors when changing the handpiece at the end of the robotic arm. Additionally, owing to the narrow oral environment, customized markers may fall outside the camera’s field of view when the robotic arm moves to certain positions.42 To tackle this problem, a dental implant robot system based on optical marker spatial registration and probe positioning strategies is designed. Zhao et al constructed a modular implant robotic system based on binocular visual navigation devices operating on the principles of visible light with “eye-to-hand” mode, allowing complete observation of markers and handpieces within the camera’s field of view, thereby ensuring greater flexibility and stability.38,58

The dental implant robotics execution system comprises hardware such as motors, force sensors, actuators, controllers, and software components to perform tasks and actions during implant surgery. The system receives commands, controls the robot’s movements and behaviors, and executes the necessary tasks and actions. Presently, research on dental implant robotic systems primarily focuses on the mechanical arm structure and drilling methods.

The majority of dental implant robotic systems directly adopt serial-linked industrial robotic arms based on the successful application of industrial robots with the same robotic arm connection.59,60,61,62 These studies not only establish implant robot platforms to validate implant accuracy and assess the influence of implant angles, depths, and diameters on initial stability but also simulate chewing processes and prepare natural root-shaped osteotomies based on volume decomposition. Presently, most dental implant robots in research employ a single robotic arm for surgery. Lai et al.62 indicated that the stability of the handpieces during surgery and real-time feedback of patient movement are crucial factors affecting the accuracy of robot-assisted implant surgery. The former requires physical feedback, while the latter necessitates visual feedback. Hence, they employed a dual-arm robotic system where the main robotic arm was equipped with multi-axis force and torque sensors for performing osteotomies and implant placement. The auxiliary arm consisted of an infrared monocular probe used for visual system positioning to address visual occlusion issues arising from changes in arm angles during surgery.

The robots mentioned above use handpieces to execute osteotomies and implant placement. However, owing to limitations in patient mouth opening, performing osteotomies and placing implants in the posterior region can be challenging. To overcome the spatial constraints during osteotomies in implant surgery, Yuan et al.63 proposed a robot system based on earlier research which is laser-assisted tooth preparation. This system involves a non-contact ultra-short pulse laser for preparing osteotomies. The preliminary findings confirmed the feasibility of robotically controlling ultra-short pulse lasers for osteotomies, introducing a novel method for a non-contact dental implant robotic system.

Position following of dental implant robotic system

It can be challenging for patients under local anesthesia to remain completely still during robot-assisted dental implant surgery.52,64,65,66,67 Any significant micromovement in the patient’s position can severely affect clinical surgical outcomes, such as surgical efficiency, implant placement accuracy compared to the planned position, and patient safety. Intraoperative movement may necessitate re-registration for certain dental implant robotic systems. In order to guarantee safety and accuracy during surgery, the robot must detect any movement in the patient’s position and promptly adjust the position of the robotic arm in real time. Yakebot uses binocular vision to monitor visual markers placed outside the patient’s mouth and at the end of the robotic arm. This captures motion information and calculates relative position errors. The robot control system utilizes preoperatively planned positions, visual and force feedback, and robot kinematic models to calculate optimal control commands for guiding the robotic arm’s micromovements and tracking the patient’s micromovements during drilling. As the osteotomies are performed to the planned depth, the robotic arm compensates for the patient’s displacement through the position following the function. The Yakebot’s visual system continuously monitors the patient’s head movement in real time and issues control commands every 0.008 s. The robotic arm is capable of following the patient’s movements with a motion servo in just 0.2 s, ensuring precise and timely positioning.

The simulation of surgeons’ tactile sensation in dental implant robotic systems

Robot-assisted dental implant surgery requires the expertise and tactile sense of a surgeon to ensure accurate implantation. Experienced surgeons can perceive bone density through the resistance they feel in their hands and adjust the force magnitude or direction accordingly. This ensures proper drilling along the planned path. However, robotic systems lack perception and control, which may result in a preference for the bone side with lower density. This can lead to inaccurate positioning compared to the planned implant position.61,62 Addressing this challenge, Li et al.68 established force-deformation compensation curves in the X, Y, and Z directions for the robot’s end-effector based on the visual and force servo systems of the autonomous dental robotic system, Yakebot. Subsequently, a corresponding force-deformation compensation strategy was formulated for this robot, thus proving the effectiveness and accuracy of force and visual servo control through in vitro experiments. The implementation of this mixed control mode, which integrates visual and force servo systems, has improved the robot’s accuracy in implantation and ability to handle complex bone structures. Based on force and visual servo control systems, Chen et al.69 have also explored the relationship between force sensing and the primary stability of implants placed using the Yakebot autonomous dental robotic system through an in vitro study. A significant correlation was found between Yakebot’s force sensing and the insertion torque of the implants. This correlation conforms to an interpretable mathematical model, which facilitates the predictable initial stability of the implants after placement.

During osteotomies with heat production (which is considered one of the leading causes of bone tissue injury), experienced surgeons could sense possible thermal exposure via their hand feeling. However, with free-handed implant placement surgery, it is challenging to perceive temperature changes during the surgical process and establish an effective temperature prediction model that relies solely on a surgeon’s tactile sense. Zhao et al.70, using the Yakebot robotic system, investigated the correlation between drilling-related mechanical data and heat production and established a clinically relevant surrogate for intraosseous temperature measurement using force/torque sensor-captured signals. They also established a real-time temperature prediction model based on real-time force sensor monitoring values. This model aims to effectively prevent the adverse effects of high temperatures on osseointegration, laying the foundation for the dental implant robotic system to autonomously control heat production and prevent bone damage during autonomous robotic implant surgery.

The innovative technologies mentioned above allow dental implant robotic systems to simulate the tactile sensation of a surgeon and even surpass the limitations of human experience. This advancement promises to address issues that free-handed implant placement techniques struggle to resolve. Moreover, this development indicates substantial progress and great potential for implantation.

Clinical research on dental implant robotic systemsClinical workflow of dental implant robotic systems

The robotic assistant dental implant surgery consists of three steps: preoperative planning, intraoperative phase, and postoperative phase (Fig. 5). For preoperative planning, it is necessary to obtain digital intraoral casts and CBCT data from the patient, which are then imported into preoperative planning software for 3D reconstruction and planning implant placement. For single or multiple tooth gaps using implant robotic systems (except Yakebot),61,62,71,72 a universal registration device (such as the U-shaped tube) must be worn on the patients’ missing tooth site using a silicone impression material preoperatively to acquire CBCT data for registration. The software performs virtual placement of implant positions based on prosthetic and biological principles of implant surgery, taking into account the bone quality of the edentulous implant site to determine the drilling sequence, insertion depth of each drill, speed, and feed rate. For single or multiple tooth implants performed using Yakebot, there is no need for preoperative CBCT imaging with markers. However, it is necessary to design surgical accessories with registration holes, brackets for attaching visual markers, and devices for assisting mouth opening and suction within the software (Yakebot Technology Co., Ltd., Beijing, China). These accessories are manufactured using 3D printing technology.

Fig. 5

Clinical workflow of robotic-assisted dental implant placement

For the intraoperative phase, the first step is preoperative registration and calibration. For Yakebot, the end-effector marker is mounted to the robotic arm, and the spatial positions are recorded under the optical tracker. The calibration plate with the positioning points is then assembled into the implant handpiece for drill tip calibration. Then, the registration probe is inserted in the registration holes of the jaw positioning plate in turn for spatial registration of the jaw marker and the jaw. Robot-assisted dental implant surgery usually does not require flapped surgery,73,74, yet bone grafting due to insufficient bone volume in a single edentulous space or cases of complete edentulism requiring alveolar ridge preparation may require elevation of flaps. For full-arch robot-assisted implant surgery, a personalized template with a positioning marker is required and should be fixed with metallic pins for undergoing an intraoperative CBCT examination, thus facilitating the robot and the jaws registration in the visual space and allowing the surgical robot to track the patient’s motion. The safe deployment of a robot from the surgical site is an essential principle for robot-assisted implant surgery. In the case of most robots, such as Yomi, the surgeon needs to hold the handpieces to control and supervise the robot’s movement in real time and stop the robotic arm’s movement in case of any accidents. With Yakebot, the entire surgery is performed under the surgeon’s supervision, and immediate instructions are sent in response to possible emergencies via a foot pedal. Additionally, the recording of the entrance and exit of the patient’s mouth ensures that the instruments would not damage the patient’s surrounding tissues. The postoperative phase aims at postoperative CBCT acquisition and accuracy measurement.

In clinical surgical practice, robots with varying levels of autonomy perform implant surgeries differently. According to the autonomy levels classified by Yang et al.6,8,33 for medical robots, commercial dental implant robotic systems (Table 2) currently operate at the level of robot assistance or task autonomy.

Table 2 The autonomous level of commercial dental implant robotics

The robot-assistance dental implant robotic systems provide haptic,75 visual or combined visual and tactile guidance during dental implant surgery.46,76,77 Throughout the procedure, surgeons must maneuver handpieces attached to the robotic guidance arm and apply light force to prepare osteotomies.62 The robotic arm constrains the 3D space of the drill as defined by the virtual plan, enabling surgeons to move the end of the mechanical arm horizontally or adjust its movement speed. However, during immediate implant placement or full-arch implant surgery, both surgeons and robots may struggle to accurately perceive poor bone quality, which should prompt adjustments at the time of implant placement. This can lead to incorrect final implant positions compared to the planned locations.

The task-autonomous dental implant robotic systems can autonomously perform partial surgical procedures, such as adjusting the position of the handpiece to the planned position and preparing the implant bed at a predetermined speed according to the pre-operative implant plan, and surgeons should send instructions, monitor the robot’s operation, and perform partial interventions as needed. For example, the Remebot77,78 requires surgeons to drag the robotic arm into and out of the mouth during surgery, and the robot automatically performs osteotomies or places implants according to planned positions under the surgeon’s surveillance. The autonomous dental implant robot system, Yakebot,73,79,80 can accurately reach the implant site and complete operations such as implant bed preparation and placement during surgery. It can be controlled by the surgeon using foot pedals and automatically stops drilling after reaching the termination position before returning to the initial position. Throughout the entire process, surgeons only need to send commands to the robot using foot pedals.

Clinical performance of robot-assisted implant surgery

Figure 6 shows the results of accuracy in vitro, in vivo, and clinical studies on robot-assisted implant surgery.20,46,48,55,62,64,67,68,69,70,71,72,75,76,77,78,79,80,81,82,83,84,85,