Senior Project - LLEAP
Our senior design project consisted of designing and manufacturing a biomechanically accurate, actuated knee joint to be integrated into an exoskeleton being developed by the Lower Limb Exoskeleton Assist Project (LLEAP), a part of the EMPOWER student association at Cal Poly, San Luis Obispo. As the human knee flexes and extends throughout gait motion, the center of rotation changes. Currently marketed exoskeletons have one point of rotation, which over constrains the knee and causes misalignment between the user and the suit [1]. Our goal was to mimic natural knee joint motion by changing the center or rotation, thus reducing misalignment and limiting power loss. We designed this knee joint for our prospective exoskeleton user: Carlo Ruggiero, a 21-year-old Cal Poly student with a complete C8 injury to his spine which resulted in loss of function and sensation from his chest and below.
Design Requirements:
- The joint must be biomechanically accurate (i.e., must match the natural roll-glide movement of the knee joint).
- The design must be lighter than the earlier LLEAP prototype (under 20 lb.).
- The knee joint must be compatible with their exoskeleton design.
- The joint must withstand the loads of the user and suit while walking.
Design Overview:
Our joint consists of a four-bar linkage tuned to mimic the natural roll-glide motion of the human knee, powered by a linear actuator (Thomson Linear LL24B040-0100LEXANNSD). The linear actuator is attached to the femur structure and actuates the joint by exerting a force on the aluminum coupler below the tibia link, resulting in a moment about the joint.
This moment causes the four-bar linkage to rotate, thereby moving the lower leg portion of the exoskeleton. The center of rotation of the joint is located at the intersection of the ACL and PCL centerlines and changes throughout the movement of the joint, mimicking the natural "roll-glide" movement of the human knee. The overall design is approximately 18 inches tall (when joint is at the 90-degree/fully retracted position), 7.5 inches deep (measuring from the front of the frontal support structure to the back of the linear actuator), and 4 inches wide (from the exterior to interior of the support structure).
Control System
To control the movement of the joint, we used an ESP32 board connected to an angular potentiometer placed on the joint (attached using a 3D printed joint shown in red in the above image). For ease of use, we connected to the ESP32 over Bluetooth Serial and controlled the knee's movement using our phones. Using the feedback from the potentiomenter, we were able to track the angular position of the knee and set it's position to specific angles.
To do this, we created a basic P controller that compared the desired angle input to the system to the current angular reading of the potentiometer (converted from an output voltage to an angular reading through a gain developed from calibration and the ESP32 datasheet). The linear actuator will then extend or retract (depending on the current and input angles) until it is within 0.5% of the input angle. A tolerance around the input angle rather than the angle itself was used because of the limited accuracy of the potentiometer and noise in the wiring system.
Results and Findings
This project created a biomechanically accurate knee joint for the Lower Limb Exoskeleton Assist Project using a four-bar linkage and linear actuator. This project was successful in force, integration, and manufacturability, but did not achieve wearability and controls requirements. For this design to be developed further, a custom or improved linear actuator must be integrated that meets all control, weight, and size requirements. This will make the device more accessible to a variety of users and improve device wearability.
The design's biomechanical accuracy was difficult to quantify, as it was not possible to determine correct link lengths based on the user's geometry as well as the amount of change in the knee's center of rotation. However, with proper imaging techniques and updated link lengths based on a defined center of rotation path for the user's knee, the correct biomechanical motion can likely be achieved in the future.
This project discovered the advantages and limitations of a four-bar linkage and linear actuator approach to creating a biomechanical knee joint. The findings of this project will allow for the development of a future design that better meets the requirements of an exoskeleton knee joint, enabling a paralyzed individual to walk again.