Virtual Constraint-based Control Method For Active Lower Limb Prostheses

The active lower limb prosthesis can provide power assist for the amputated patients.Compared with the passive prosthesis,active prosthesis can produce more natural gait,reduce the wearer’s energy consumption and achieve more motion patterns.With the development of science and technology,the active prosthesis is gradually replacing the passive prosthesis and has become the focus of research.This thesis mainly includes the recognition of the terrain in front of the user,building dynamic model of prosthesis and the control method of prosthesis.The main contents of this thesis are as follows:Firstly,in order to provide environment information to the active lower limb prosthesis,a wearable terrain recognition system is designed.The 2D lidar installed in the waist is used to collect the terrain data.Then the linear feature of the terrain data is extracted using agglomerative hierarchical clustering algorithm.Based on the results of linear feature extraction,the terrain is recognized using non-deterministic finite automaton.Four kinds of terrains is recognized including level ground to up/down stair and level ground to up/down slope.In addition,the agglomerative hierarchical clustering algorithm and hough transform algorithm are compared.The superiority of the algorithm adopted in this thesis is verified.The problem of terrain recognition for the active lower limb prosthesis is effectively solved.Secondly,the VICON MX gait information collection platform is used to collect the human gait information.The joint angle information of walking on five terrain,including the flat,up / down stairs and on / off ramp,is collected and analyzed.Thus,the importance of terrain recognition for prosthesis control is proved.And the joint angle information are also used for establishing virtual constraint of active lower limb prosthesis.Thirdly,the prosthesis is simplified into three parts: the thigh,the shank and the circular foot.Circular foot is used to capture the center of plantar pressure movement.Compared with point foot model,the dynamic characteristics of circular foot model are closer to the human lower limb.Based on the simplified model,the dynamic model of human lower limb is established by solving the Lagrange equation.Finally,aiming at the problem of active lower limb prosthesis and human body coordination control,a virtual constraint-based control method is applied to control the active lower limb prosthesis.In order to further improve the convergence speed of the prosthesis control system,a finite-time stabilization control method is proposed to replace the traditional feedback linearization control.The experimental results show that the finite-time stabilization control method can effectively control the lower limb prosthesis.Compared with traditional feedback linearization control,finite-time stabilization control can track the virtual constraints more accurately and have a faster convergence speed.