The global surgical robotics market was valued around $8 billion in 2022 and is expected to continue growing at around 16.6% CAGR in the next decade, reaching $18 billion by 2027 and $36 billion by 2032. The market is dominated by Intuitive and Stryker who in total hold around 75% of the current market. J&J MedTech and Medtronic plan to tap into this growing market by introducing OTTAVA and Hugo RAS product lines respectively. The major fields where robotic surgery has either been adopted or are being adopted are general surgery, cardiology, orthopedics, neurosurgery and vascular surgery.
Robotics surgery offers several benefits to surgeons as compared to the traditional forms of surgery such as greater dexterity, superior maneuverability, greater accuracy, repeatability and control. However, the challenges still remain including surgical errors and other complications. In this article, we will examine how Gen AI technologies can improve the accuracy, control, repeatability and dexterity of surgical robots.
Gen AI technologies are generally associated with the creation of texts, images and videos. These technologies are based on what are called Large Language Models(LLMs) which are models trained on large set of data consisting of mostly texts and images. The next step in the evolution is what are called VLAMs that stand for Vision-Language-Action Models. These models are not only trained on text and images, but they also feed on data present in the physical world (like position of an object), readings of different sensors, rotation and movement of moving parts and other cues than a human mind generally observes while taking an “action” of following a command. These models are also called “multimodal models”. These models are much more powerful than LLMs and find direct use in Robotics. There are several instances of such models already available on the market. For example, Covariant has created a model called RFM-1.
VLAMs are fundamentally changing the Robotics market and acting as a “Brain” for what were previously mostly passive command driven Robots. VLAMs grant a sense of common knowledge to the Robots. Another contribution of VLAMs towards Robotics is “in-context learning”. Robots can now be programmed just by giving commands on text prompts rather than requiring elaborate reprogramming. Such VLAMs powered robots are already finding usage in warehouses and factory floors.
As usual, MedTech is always a laggard in adopting the latest innovations. However, to stay competitive, they will have to sooner or later integrate VLAMs in Robotics Surgery. Although Robotics surgery will always be under the guidance of a “human” brain in the foreseeable future, making the Robots smarter will do no harm. Using VLAMs, RAS (Robot Assisted Surgery) systems can automatically take action based on movement of patients in a OR, get trained on how to access difficult to reach parts, take voice and text-based commands to do routine procedures and do other wonders. This field is still untouched and the companies trying to enter the Robotics Surgery market will do great good to themselves by integrating VLAMs in the development of their RAS platforms.
