G. Quiroz

State and parameter estimation of an anaerobic digester model

Wastewater treatment is a current technological and scientific challenge due to the requirements of the high variety of wastewater coming from industries and municipal residuals. In particular, anaerobic digestion process is a useful approach for treatment of wastewater with high load of organic matter. The need of implementing treatment plants for anaerobic digestion has open many research opportunities about modeling and control of this complex biochemical process. In particular, this contribution deals with the lack of variables information from available sensors. We propose a state and parameter estimation of a nonlinear dynamical model of anaerobic digestion; specifically, the proposed approach is able to estimate the states: acidogenic and methanogenic bacteria concentrations, as well as the parameters: maximum bacterial growth rates for acidogenesis and methanization, and the hydrodynamic regimen. The estimation approach is based on adaptive observers, and it provides a reliable tool towards the control of anaerobic digesters.

Inverse Kinematics Estimation Based on Smart Optimization for Biomechanical Linkages

To obtain an analytical solution of inverse kinematics in complex biomechanic linkages is a difficult, even impossible problem. This could be because the kinematic redundancy and the large number of degrees of freedom that it involves. EvoNorm is a smart optimization algorithm that can be useful to solve this kind of problems. For this reason, in this work a numerical solution of the inverse kinematics problem for biomechanical linkages is proposed. It is computed using EvoNorm by minimizing a fitness function based on the joints position. Two prototypes are presented as study cases: a hand and a lower-limb exoskeleton. Results illustrate the performance of the proposed methodology.

Hybrid BCI approach to control an artificial tibio-femoral joint

Brain-Computer Interfaces (BCIs) for disabled people should allow them to use their remaining functionalities as control possibilities. BCIs connect the brain with external devices to perform the volition or intent of movement, regardless if that individual is unable to perform the task due to body impairments. In this work we fuse electromyographic (EMG) with electroencephalographic (EEG) activity in a framework called “Hybrid-BCI” (hBCI) approach to control the movement of a simulated tibio-femoral joint. Two mathematical models of a tibio-femoral joint are used to emulate the kinematic and dynamic behavior of the knee. The interest is to reproduce different velocities of the human gait cycle. The EEG signals are used to classify the user intent, which are the velocity changes, meanwhile the superficial EMG signals are used to estimate the amplitude of such intent. A multi-level controller is used to solve the trajectory tracking problem involved. The lower level consists of an individual controller for each model, it solves the tracking of the desired trajectory even considering different velocities of the human gait cycle. The mid-level uses a combination of a logical operator and a finite state machine for the switching between models. Finally, the highest level consists in a support vector machine to classify the desired activity.

An Embedded Generation of Fuzzy Rules Implemented in a Remote Car Driving

The aim of the present research and development work, is to show how to implement a embedded Generation of Fuzzy Rules.

Establishing a Simplified Functional Relationship between EMG Signals and Actuation Signals Using Artificial Neural Networks

Using EMG signals as control signals has been a widely accepted option in the last decades. Using a wide array of techniques, EMG signals can be used in a variety of practical ways, from prostethics to exoesqueletons, however a concrete functional relationship between EMG signals and the dynamic and kinematic aspects of the upper limbs has not been established. Nowadays, almost every device that uses EMG signals uses them for classification purposes. In this work, we employ Fourier analysis in conjunction with other signal processing tools to treat the EMG signal, the treated signal is then used as an input of an artificial neural network in order to establish a simplified functional relationship between EMG and the upper limbs. We also employed other traditional signal processing methods for comparison purposes.