Self Adjustable Intelligent Knee Mimicking the Real Human Gait-Cycle Using Machine Learning

This research focuses on developing a novel control system technique for electronically controlled knee joints to help amputees achieve the actual human gait cycle without facing any difficulties. The proposed technique controls the movement of electronically controlled microprocessor knee intelligently by acquiring sensory data from the amputee's functional leg. As a result, amputee will be able to achieve a gait cycle like actual human gait cycle. This research will prove to be a revolution in the field of intelligent prosthesis as the patient will not be bound to limited modes of walking, running and jogging etc. rather he will be able to achieve any stance like walking slowly, or fast and running or jogging. Real-time sensory data is acquired from different points of the functional leg and then used to mimic the behavior of functional leg on to the prosthetic knee. The acquired sensory data includes angular movement, acceleration and the forces exerted on the ground. This sensory data is then passed to a machine learning algorithm for optimizing results and controlling prosthetic knee through beagle-bone microcontroller. The results are simulated in ROS software on a two degree of freedom leg with three links and three revolute joints. The results achieved are very promising, as the behavior of functional leg is efficiently mimicked up to 90%.