AbstractThere were 1.6 million people who suffered from limb loss in the U.S in 2005 alone. Amputation changes one's body function and life dramatically, especially lower limb losses. Prostheses can largely save the situation. Considerable effort is currently being expended on developing intelligent powered lower-limb prostheses. These prostheses require patient tuning and validation during the developing process. Present prosthesis testings are mainly based on amputees and prosthetists qualitative feedback, which is inconsistent and unsafe. It also limits the test conditions. Additionally, the recruitment of amputees is hard. This work seeks approaches to test lower limb prostheses to replace human testing.
An investigation of applying Hardware-in-the-Loop (HIL) on testing prostheses
is presented. HIL testing has been used successfully for a number of years on a
wide range of applications. A HIL testing system has groundbreaking potential in
prostheses testing to investigate the nature of human learning of walking. However, any delay in a HIL system will cause an increase in energy. The stiff ground contact discontinuity is hard to compensate. We investigate the effect of introducing nonlinearity and discontinuity into a HIL system by comparing three types of Spring Mass System (SMS). Lead Compensation (LC) and Horiuchi Compensation (HC) are used and compared. It is concluded that the actuation system delay frequency should be 20 times greater than the system natural frequency in order to keep the system simulation stable, which is hard to realise.
Then a novel approach to test lower limb prostheses with the development of a
Hydraulic Gait Simulator (HGS) and control strategy is presented. A gait simulator
testing of lower-limb prostheses has the advantage of (i) removing humans from
early-stage experimental testing (ii) gathering objective quantitative measurements and (iii) and permits the inspection of joint reaction forces (non-achievable without having instrumented joints in the amputees). The approach uses a leg robot with a prosthesis foot to achieve required Ground Reaction Force (GRF) of human use. The kinematics, kinetics of test prosthesis and the robot leg energy are used to evaluate the performance of the test prosthesis. It is the first robot-based lower limb prosthesis testing method that tests prostheses by generating walking gaits and performing a quantied evaluation. To address the GRF control, an Extended Iterative Learning Control (EILC) algorithm is derived. The process of achieving the required GRF can be seen as a learning process from a test prosthesis user. The algorithm is validated both in simulation and experiment. It is found effective on stationary ground, moving ground with passive ankle prostheses and active ankle prostheses. An example of testing a passive ankle prosthesis is presented in the end. The HGS is controlled with the proposed EILC to replicate human walking GRF. Different settings of the prostheses are used. By analysing the generated gaits, the HGS has been proven to be able to identify small changes in the prosthesis. The
prosthesis is observed to give insuffcient power in the Powered Plantar
flexion (PP) phase and this finding agrees with traditional human testing results, demonstrating the effectiveness of the proposed test approach in testing a lower leg prosthesis.
|Date of Award||22 Jul 2020|
|Supervisor||Pejman Iravani (Supervisor), Andrew Plummer (Supervisor) & Min Pan (Supervisor)|
- Iterative learning control
- motion control