Project Details
Projekt Print View

Robust phase-based control of prosthetic feet and biologically inspired joint coupling

Subject Area Automation, Mechatronics, Control Systems, Intelligent Technical Systems, Robotics
Term from 2018 to 2023
Project identifier Deutsche Forschungsgemeinschaft (DFG) - Project number 394182789
 
Final Report Year 2023

Final Report Abstract

In this project, we investigated whether asymmetries and effort during level walking can be positively influenced using a novel prosthetic prototype that implements next to a monoarticular actuator at the ankle joint a biarticular actuator that is coupled to the thigh. Additionally, the goal was to develop a method that enables continuous interaction with the prosthesis for level walking, stair climbing, and transitions between these gait modes. To address both goals, a motion dataset (ground reaction forces, joint positions, motion data of limb segments through inertial measurement units, muscle activity) was recorded for a walking track with stairs at three different stair heights. The data shows that transition phases between gait modes exist, which significantly differ from level walking and stair ambulation. Furthermore, stair ascent requires up to 13 times the positive net leg work and stair descent requires up to 13 times the negative net leg work, compared to level walking. This indicates that active prosthetic components for the lower extremities should consider the specific characteristics of stair ascent and descent. Using the data, a prosthetic foot with mono- and biarticular actuation and control was designed. After initial tests with non-amputee subjects, the interaction between mono- and biarticular actuation was examined in individuals with unilateral transtibial amputation. As hypothesized, walking effort could be reduced (9% with a 22% level of biarticular support) and gait asymmetries could be influenced using the biarticular actuator instead of the monoarticular one. The highly individual gait behavior of the users suggests that a more pronounced customization of support throughout the gait cycle is necessary for both the overall behavior of the active ankle joint and the distribution between mono- and biarticular actuators. Furthermore, the dataset was used to develop solutions for gait recognition, phase detection, and stair slope identification. It was found that the low-dimensional concepts based on phase portraits outlined in the proposal do not allow for unambiguous determination. Therefore, beyond the scope of the proposal, higher-dimensional machine learning methods were investigated for their suitability. Kinematic measurements from inertial measurement units (IMUs) were used as input for an Artificial Neural Network (ANN). For validation on non-amputee subjects in various environments, a mobile measurement setup (FRIMU) with IMUs and ground contact sensors was developed. The results demonstrate that gait recognition, phase detection, and stair slope identification are feasible. Moreover, we see great potential for transferring these findings to thematically related fields of wearable robotics.

Publications

 
 

Additional Information

Textvergrößerung und Kontrastanpassung