Researchers at the Harvard John A. Paulson School of Engineering and Applied Sciences have developed a mathematical framework for optimizing the design of rolling contact joints, a class of robotic joints composed of curved rolling surfaces connected by flexible elements. The method enables designers to adjust the geometry of all joint components simultaneously to meet specific force and motion requirements, rather than relying primarily on software-based control strategies.
The approach was detailed in a peer-reviewed study published in the Proceedings of the National Academy of Sciences. The framework allows joint geometry to be tailored to a desired mechanical output, such as force transmission at different points along a motion path, by embedding task requirements directly into the physical design of the joint.
Rolling contact joints are inspired by biological systems such as the human knee, which combines rotation with rolling and sliding motion. In robotics, such joints can offer flexibility, low friction, and resistance to wear, but their design has traditionally been limited to simple circular geometries. The new method supports noncircular and irregular shapes that can follow complex trajectories while delivering targeted mechanical performance.
To demonstrate the framework, the research team constructed two prototypes: a knee-like joint and a two-finger robotic gripper. In the knee joint experiment, the researchers mapped the average motion path of a human knee and used the optimization method to design a joint that closely reproduced that movement. When compared with conventional bearing-based mechanisms commonly used in knee-assist devices and exoskeletons, the optimized joint reduced misalignment by 99 percent.
In a second demonstration, the team applied the method to a robotic gripper designed to handle objects of varying sizes. By optimizing the rolling contact joints to adjust force delivery along the fingers’ motion, the gripper was able to support more than three times the load of a comparable gripper built with standard circular joints and pulleys, using the same actuator input.
According to the researchers, the design framework enables closer coupling between a robot’s physical structure and its intended task, potentially reducing the need for complex control systems and allowing for smaller actuators. The method is applicable across a range of domains, including task-specific robots, assistive and rehabilitative devices, and studies of animal and human biomechanics.
Photo credit: Wood Lab / Harvard SEAS
