Article Preview
Top1. Introduction
Identification of technical processes for analysis, optimization and control is a major challenge. This focus also includes the use of identification methods and applications for the biotechnology sector. In this project identification in particular plays an important role in enabling an interface between the brain and the control of movement based on data from peripheral nerves.
So, the identification of motion- and sensory feedback-based action potentials in peripheral nerves is a great challenge in medical technology. It is the prerequisite for applications like prosthesis control or limb stimulation. Based on the acquisition of action potentials, the identification process correlates physiological and motion-based parameters to match movement trajectories and the corresponding action potentials.
The identification method used in this context is based on the continuous mode symbiotic cycle, combining a physical system, a simulation system and an agent-based machine learning system. As an example, a data-driven method to create a human readable model without using presampled data is presented. All components in the system interact in a symbiotic way. The result of each component is used as an input by the others and vice versa.
First of all, the prototype of biosignal acquisition and identification system using a multistage agent-based solution builder identification method is introduced (Klinger and Klauke, 2013) and then the closed-loop identification method, implemented using a symbiotic continuous system (Aydt, Turner, Cai and Low, 2008; Aydt, Turner, Cai and Low, 2009) is presented. The prototype is acting as the physical target system in the symbiotic cycle, presented subsequently. This paper focuses on the interaction between the identification method, based on a data driven approach and its verification. We present the closed-loop identification method, implemented using a symbiotic continuous system (Aydt, Turner, Cai and Low, 2008; Aydt, Turner, Cai and Low, 2009), consisting of a robotic based trajectory generation, the nerve simulation and an agent-based machine learning system. We introduce the model generation process and show the closed-loop verification approach of the identification method.