Session: ASME Undergraduate Student Design Expo

Paper Number: 175106

Compliant Rolling Contact Joint for Bipedal Robots 

Compliant mechanisms differ fundamentally from rigid mechanisms. While rigid mechanisms rely on discrete rigid links and joints, compliant mechanisms use the elastic deformation of flexible members and joints to transmit motion and force. This nature-inspired approach eliminates the need for traditional bearings and hinges, reducing part count, assembly requirements, and friction losses. Enabled by advances in additive manufacturing, compliant mechanisms can be fabricated as a single piece, which improves reliability, reduces wear, and expands design possibilities. Their main challenge lies in the complexity of modeling nonlinear deformations. Despite this, they have found wide application in robotic locomotion, microelectromechanical systems (MEMS), biomedical devices, and robotic grippers.

In this study, we designed and built a bipedal robot featuring a compliant rolling contact joint that mimics the polycentric behavior of the human knee—difficult to replicate with rigid-link joints. Each leg was 3D printed as a single piece: the upper and lower limbs were fabricated from polylactic acid (PLA), while the rolling contact joints were printed using thermoplastic polyurethane (TPU) for flexibility. Each leg is actuated by three 25 kg servo motors. A Raspberry Pi, battery pack, and camera were integrated into the torso, and wireless IMU sensors were attached to both the upper and lower legs.

The gait cycle was implemented in two distinct ways. First, we collected human motion data using the Movella XSens Awinda suit equipped with 14 IMU sensors, capturing static and dynamic walking at slow and fast speeds. This dataset was used to replicate human-like gait, allowing the robot to closely mimic natural walking motion. Second, we applied the Zero Moment Point (ZMP) method. ZMP is a widely used criterion for gait stability in bipedal robots, defined as the point on the ground where the net moment of inertial and gravitational forces equals zero. By keeping the ZMP within the support polygon of the foot, the robot maintains dynamic balance during walking. This method enables predictive control of stability and allows the robot to adapt step timing and placement to prevent falling. Using this approach, we achieved stable walking at slow speeds, even in conditions where human-data-driven control produced less consistent results. Comparisons of the two methods show that the human-data gait enables lifelike motion, while ZMP provides reliable balance and stability control.

This design demonstrates how compliant bistable joints can simplify robotic leg construction while improving motion fidelity. By eliminating traditional hinges and incorporating rolling contact joints, the robot achieves reduced part count, lower friction losses, and improved durability. The ability to replicate natural human gait using motion data, combined with stability through ZMP control, shows the potential of this approach for next-generation robotic systems. Such a design could be applied in assistive walking devices, rehabilitation platforms, and aging technologies where safe, stable, and lifelike motion is essential. It may also serve as a foundation for entertainment robots and humanoid systems requiring lightweight and reliable legged locomotion.

 

Presenting Author: Regina Martinez Moctezuma Kennesaw State University

Presenting Author Biography: Regina Martinez is a mechatronics and robotics undergraduate research assistant at Kennesaw State University.

Authors:

Regina Martinez Moctezuma Kennesaw State University
Sabrina Scarpinato Kennesaw State University
Andrew Parella kennesaw state university
Deniz Kerimoglu Georgia Institute of Technology
Daniel Goldman Georgia Institute of technology
Ayse Tekes Kennesaw State University