Session: 06-10-01: Robotics, Rehabilitation

Paper Number: 144981

144981 - Intention-Predicted Motion Planning of an Assistive Bionic Joint Knee Brace for Knee Rehabilitation 

Introduction:

Chronic knee pain, also medically known as knee osteoarthritis (KOA), is slowly crippling adults in the U.S. Recent statistics from the CDC report that as many as 1 in 4 adults in the U.S. suffers from osteoarthritis, or approximately 30 million adults. There are many lower limb exoskeletons (LLEs) on the market today that are available to help alleviate or rehabilitate patients suffering from KOA. However, the most affordable LLEs are passive systems that lack movement adaptability or adequate torque generation, and active LLEs that do embody those capabilities are notoriously complex systems both in size and cost. In this work, we improve the efficiency and adaptability of a wearable exoskeleton device called “assistive bionic knee joint” brace (ABJ) through the implementation of intention-based adaptive electromyography (EMG) control on its fluidic muscles. This helps to perform assist-as-needed (AAN) support while walking with the knee brace and optimize the coordination between the brace and the user’s natural gait using intension-based model predictive control, thereby promoting improved seamless and intuitive lower extremity in rehabilitation with KOA and other knee injuries.

Contribution of the work:

This work aims to enhance the functionality of assistive bionic knee joint (ABJ) braces by implementing an advanced adaptive mobility algorithm. The new algorithm will leverage 3D motion analysis, Electromyography (EMG) sensors, piezoresistive force sensing, proportional pressure valves to sense user intention, collect real-time gait data and record the responsiveness of the fluidic muscles. By incorporating model predictive analysis, the algorithm will assist knee motion by user intention, potentially surpassing traditional exoskeleton systems in mimicking natural gait. With its closed-loop adaptive control system, the ABJ prototype not only offers superior movement imitation but also provides assist-as-needed (AAN) capabilities for effective knee rehabilitation. This work represents a significant advancement in assistive technology, promising improved gait efficiency and enhanced rehabilitation outcomes for patients.

Methodology:

In this work, we first conduct gait analysis during normal walking without an ABJ using a 3D motion analysis kit. Then human participants wear the ABJ exo-suit with all sensing units and actuated fluidic muscles based on the gait trajectory for open-loop experimental testing. EMG data from the participant’s thigh, force/pressure data on the ABJ brace for intention detection, and the 3D leg motion trajectory are captured and recorded. We then conduct offline simulation based on the sensing data, followed by the design and calculation of the adaptive control law and tuning of fluidic muscle variables. The data-driven adaptive algorithm from experimental testing will be implemented into the ABJ control system. Rigorous testing and continuous parameter tuning are embedded to produce an adequate method that can predict the torque required to keep the user moving effectively and comfortably.

With the help of intention detection and adaptive torque implementation, the artificial fluidic muscles have the potential to outperform current traditional exoskeleton systems in terms of mimicking the user’s natural gait with comfort and assist-as-needed (AAN) capabilities.

Preliminary results and conclusions:

We have successfully fabricated the piezoresistive pressure sensor with low-cost conductive fabric and pressure velostats in the lab. Additionally, all other sensing units were readily acquired off the shelf. In the preliminary phase, we configure an Arduino microcontroller with Simulink, which allows for initial testing of various components and sensors, and the creation of a real-time close-loop control using MATLAB. Circuits are constructed to interface with power, sensors and hardware through the microcontroller. Successfully configuring the hardware to communicate with the microcontroller laid a solid foundation for future test data collection and enabling the implementation of algorithm-based controls of the knee brace.

Presenting Author: Kiet Duong San Jose State University

Presenting Author Biography: Kiet Duong is a Graduate Research Assistant in Mechanical Engineering Department at San Jose State University. His research interest is in sensors, adaptive control, and robotics, as well as design optimization.

Authors:

Kiet Duong San Jose State University
Sidharth Swaminathan IntelliScience Training Institute
Rishit Agrawal IntelliScience Training Institute
Syed Zaidi IntelliScience Training Institute
Vimal Viswanathan San Jose State University
Li Jin San Jose State University
Lin Jiang San Jose State University