Path Planning of Surgical Robot for Minimally Invasive Surgery on Riemannian Manifold

Inside the human body, the space available for a surgeon to work is often a tight, curved chamber. When performing minimally invasive surgery, a doctor inserts long, slender tools through a small opening in the abdominal wall. These tools must navigate around delicate organs without touching them, all while pivoting on that single, fixed entry point. This constraint creates a difficult geometric puzzle: the robot holding the tool cannot simply move in a straight line, because the tool's tip might collide with an organ, or the robot's own joints might hit their physical limits. For decades, surgeons have guided these robotic arms by hand, but the shortage of skilled robotic surgeons has driven a push toward machines that can plan their own safe paths. The challenge is to find a route that is not only safe from collisions but also gentle on the robot's mechanics and the patient's body wall.

A team of researchers at The University of Tokyo and the National Institute of Informatics has developed a new way to solve this navigation problem. Instead of trying to map the path in the three-dimensional space where the robot's tip moves, they transformed the problem into a different kind of space that describes the angles of the robot's joints. They treated this space of joint angles as a curved surface, similar to how a map of the Earth is a flat representation of a sphere. By doing this, they could use mathematical tools designed for curved surfaces to find the smoothest, safest route for the robot. Their method treats the robot's joints as if they are walking on a landscape where obstacles are hills to be avoided. They found that for the specific shapes of internal organs, this "landscape" of joint angles is surprisingly simple to navigate, allowing the robot to calculate a safe path with less strain on its motors than traditional methods.

The researchers tested their idea using a robotic arm designed for surgery, which has five moving joints and a tool attached to the end. In a typical scenario, the robot must move its tool from the entry point on the body wall to a target organ, such as a gallbladder or a bladder, without hitting the surrounding tissue. The team created a digital simulation of a human abdomen, using a concave shape to represent the body cavity and placing a sphere inside to mimic an organ like a gallbladder. They then ran two different planning strategies. The first was a standard approach that plans the path in the physical space where the tool tip moves. The second was their new method, which planned the path by looking at the angles of the robot's joints.

In the standard approach, the computer tries to draw a line through the 3D space of the body cavity. Because the body cavity is curved and the organs are complex, early approaches using voxel or mesh approximations for obstacle projection often suffered from high memory demands or coarse collision checks. The researchers' new method flipped the perspective. They realized that when the robot's joints move, the "forbidden" areas created by the organs look different in the space of joint angles. Specifically, they found that the complex, curved shapes of the organs, which are concave in the physical body, often appear as simple, convex shapes in the space of joint angles. A convex shape is like a smooth hill that curves outward, whereas a concave shape is like a cave or a bowl that curves inward. This distinction is crucial because finding the shortest path around a smooth hill is mathematically much easier and faster than finding a path around a cave.

By converting the problem into this joint-angle space, the researchers could use a fast, efficient algorithm to find the best route. They defined a cost for every possible movement, penalizing paths that brought the robot too close to its physical limits or too close to the simulated organs. They then used a sampling technique to build a map of safe points and connected them with the lowest-cost paths using Dijkstra's algorithm. When they compared the results, the new method produced a trajectory that required significantly less movement from the robot's joints. In their tests, the average angle of the tool relative to the body wall was about 31.8 degrees with the new method, compared to 39.5 degrees with the traditional method. The maximum angle reached was 40.3 degrees versus 47.7 degrees. These smaller angles mean the robot is not twisting as sharply against the patient's body wall, which reduces stress on the entry port and lowers the risk of tissue damage.

The study also showed that the new method was computationally efficient. Because the obstacles in the joint-angle space could be treated as simple convex shapes, the computer did not need to perform a massive, exhaustive search to find the solution. Instead, it could use a "greedy" approach to find the closest point on the obstacle boundary within the graph, which was then used as a component in the broader Dijkstra search. This is similar to how a hiker might walk up a smooth hill to find the summit, knowing that the path is clear, rather than having to explore every nook and cranny of a complex cave system. The researchers validated these findings through experiments using a robotic arm and a mimic bladder, confirming that the calculated paths were indeed smoother and required less joint motion.

This work suggests a promising direction for the future of robotic surgery. While the current study focused on specific, simplified organ shapes, the results indicate that for many common surgical tasks, the complex geometry of the human body can be simplified when viewed through the lens of the robot's own mechanics. The researchers noted that their method is particularly well-suited for situations where the organs can be approximated as combinations of convex surfaces. They acknowledged that more complex obstacles might require further refinement of the conditions needed for this simplification to work. However, the core finding is clear: by changing the way the problem is viewed, from the space of the body to the space of the robot's joints, it becomes possible to generate safer, more efficient paths. This approach could eventually allow surgical robots to operate with a higher degree of autonomy, reducing the workload on surgeons and making minimally invasive procedures even safer for patients.