Bio-inspired adaptive feedback error learning architecture for motor control

This study proposes an adaptive control architecture based on an accurate regression method called Locally Weighted Projection Regression (LWPR) and on a bio-inspired module, such as a cerebellar-like engine. This hybrid architecture takes full advantage of the machine learning module (LWPR kernel) to abstract an optimized representation of the sensorimotor space while the cerebellar component integrates this to generate corrective terms in the framework of a control task. Furthermore, we illustrate how the use of a simple adaptive error feedback term allows to use the proposed architecture even in the absence of an accurate analytic reference model. The presented approach achieves an accurate control with low gain corrective terms (for compliant control schemes). We evaluate the contribution of the different components of the proposed scheme comparing the obtained performance with alternative approaches. Then, we show that the presented architecture can be used for accurate manipulation of different objects when their physical properties are not directly known by the controller. We evaluate how the scheme scales for simulated plants of high Degrees of Freedom (7-DOFs).     

Computational nature of human adaptive control during learning of reaching movements in force fields

Learning to make reaching movements in force fields was used as a paradigm to explore the system architecture of the biological adaptive controller. We compared the performance of a number of candidate control systems that acted on a model of the neuromuscular system of the human arm and asked how well the dynamics of the candidate system compared with the movement characteristics of 16 subjects. We found that control via a supra-spinal system that utilized an adaptive inverse model resulted in dynamics that were similar to that observed in our subjects, but lacked essential characteristics. These characteristics pointed to a different architecture where descending commands were influenced by an adaptive forward model. However, we found that control via a forward model alone also resulted in dynamics that did not match the behavior of the human arm. We considered a third control architecture where a forward model was used in conjunction with an inverse model and found that the resulting dynamics were remarkably similar to that observed in the experimental data. The essential property of this control architecture was that it predicted a complex pattern of near-discontinuities in hand trajectory in the novel force field. A nearly identical pattern was observed in our subjects, suggesting that generation of descending motor commands was likely through a control system architecture that included both adaptive forward and inverse models. We found that as subjects learned to make reaching movements, adaptation rates for the forward and inverse models could be independently estimated and the resulting changes in performance of subjects from movement to movement could be accurately accounted for. Results suggested that the adaptation of the forward model played a dominant role in the motor learning of subjects. After a period of consolidation, the rates of adaptation in the internal models were significantly larger than those observed before the memory had consolidated. This suggested that consolidation of motor memory coincided with freeing of certain computational resources for subsequent learning. Received: 01 January 1998 / Accepted in revised form: 26 January 1999     

A Controller Design Method Based on a Neural Network for an Outdoor Mobile Robot

A wheeled mobile mechanism with a passive and/or active linkage mechanism for travel in rough terrain is developed and evaluated. In our previous research, we developed a switching controller system for wheeled mobile robots in rough terrain. This system consists of two sub-systems: an environment recognition system using a self-organizing map and an adjusted control system using a neural network. In this paper, we propose a new controller design method based on a neural network. The proposed method involves three kinds of controllers: an elementary controller, adjusted controllers, and simplified controllers. In the experiments, our proposed method results in less oscillatory motion in rough terrain and performs better than a well tuned PID controller does.     

Simplified off-gas analyses in animal cell cultures for process monitoring and control purposes

Batch-to-batch reproducibility of animal cell cultures can significantly be enhanced using process control procedures. Most informative signals for advanced process control can be derived from the volume fractions of oxygen and carbon dioxide in the vent line of the reactors. Here we employed simple low-cost sensors, previously not considered for off-gas analysis at a laboratory-scale cell cultures, and compared them with a simultaneously used quadrupole mass spectrometer, i.e., the standard equipment. A decisive advantage is that the sensors did not need any calibration and are easy to use. We show that monitoring and advanced control of cell cultures can significantly be simplified using the devices tested here and that the same batch-to-batch reproducibility can be obtained with much less effort than before.     

On-line control of atracurium induced muscle relaxation

We described recently a microcomputer system capable of controlling muscle relaxation during surgical procedures; the system was tested and evaluated in 42 clinical trials involving the use of the muscle relaxant, d-tubucurarine. The advent of new non-depolarizing neuromuscular blocking drugs with significant clinical advantages makes it essential that any automatic control system for muscle relaxation can also be used with such drugs, and benefit from their improved properties. This paper describes a series of 22 clinical trials in which our controller was used successfully to control muscle relaxation using atracurium. We also investigated an alternative control strategy, taking advantage of the rapid elimination of atracurium from the body.     

An architecture for a shop-floor controller using colored Petri nets

In a dynamic and flexible manufacturing environment, a shop-floor controller must be designed so that it automatically (or with minimum human intervention) and quickly responds to the changes (e.g., in part type or part routing) in the system. Such a performance may be achieved provided that the controller is simple and sufficiently general in its scope of application. In this article, we present an architecture for such a shop-floor controller. The architecture is based on colored Petri nets with ordered colored sets and structured input and output functions.     

Control and optimization of apheresis procedures in a COBE 2997 cell separator

To obtain more efficient operation of a COBE Model 2997 clinical cell separator using either a Single Stage II (SS II) or a Dual Stage separation chamber, modifications were made to allow complete computer control. Product cell density was detected using an optical sensor and controlled by automatic feedback through a microcomputer interface. Control was accomplished by automatically adjusting the red blood cell (RBC) and plasma product flow rates using a proportional-integral (PI) algorithm. Results show that, using either chamber, the product cell density can be maintained at a preselected value for extended periods of time without operator intervention. This system allowed investigation of optimal operating regions for plateletpheresis and leukapheresis procedures. The effects of centrifuge rpm and controller set point on centrifuge operation were investigated using a second order factorial experimental design. Theoretical significance of model parameters was assessed with the aid of a hindered settling model and simple reasoning about the interface position relative to the collection port. The results suggest that, in either chamber, the optimum operating region for plateletpheresis procedures occurs at moderate controller set points and high centrifuge rpm. The resultant operating efficiency and product purity values are approximately 63 percent and 0.65 respectively in the SS II chamber and approximately 70 percent and 0.70 respectively in the Dual Chamber. In the SS II, the optimum operating region for leukapheresis procedures occurred at high controller set point values for any centrifuge rpm above 1200 with an operating efficiency near 100 percent. However, in the Dual Chamber, the optimum operating region for leukapheresis procedures occurred at high controller set points and high centrifuge rpm's, again providing an operating efficiency near 100 percent.     

New techniques in rapid viral diagnosis

Abstract The development of new diagnostic techniques in immunology and molecular biology during the last two decades has opened up new possibilities for rapid viral diagnosis. Solid phase immunoassays for antigen and antibody detection are now widely used in diagnostic settings. Several novel techniques have been introduced and have led to commercially available tests. Diagnostic methods using nucleic acid amplification procedures are already applied in research laboratories and will be commercialized soon. Biosensor-based diagnostic techniques have the potential of generating a result nearly instantaneously and it has become possible to monitor kinetic processes. Automatization and simplified procedures are needed to allow diagnostic tests to be performed soon after the sample has been obtained from the patient. In order to evaluate the new procedures and avoid false results, rigorous quality control in diagnostic virology will have to be instituted.     

Towards Active Tracking of Beating Heart Motion in the Presence of Arrhythmia for Robotic Assisted Beating Heart Surgery

In robotic assisted beating heart surgery, the control architecture for heart motion tracking has stringent requirements in terms of bandwidth of the motion that needs to be tracked. In order to achieve sufficient tracking accuracy, feed-forward control algorithms, which rely on estimations of upcoming heart motion, have been proposed in the literature. However, performance of these feed-forward motion control algorithms under heart rhythm variations is an important concern. In their past work, the authors have demonstrated the effectiveness of a receding horizon model predictive control-based algorithm, which used generalized adaptive predictors, under constant and slowly varying heart rate conditions. This paper extends these studies to the case when the heart motion statistics change abruptly and significantly, such as during arrhythmias. A feasibility study is carried out to assess the motion tracking capabilities of the adaptive algorithms in the occurrence of arrhythmia during beating heart surgery. Specifically, the tracking performance of the algorithms is evaluated on prerecorded motion data, which is collected in vivo and includes heart rhythm irregularities. The algorithms are tested using both simulations and bench experiments on a three degree-of-freedom robotic test bed. They are also compared with a position-plus-derivative controller as well as a receding horizon model predictive controller that employs an extended Kalman filter algorithm for predicting future heart motion.     

A model of internal control may improve the response time of an automatic arterial pressure controller

A simplified model for the arterial pressure control system was implemented on a personal computer using Matlab Simulink. Model responses to variations of systemic vascular resistance were comparable to those predicted by physiology. Computer simulation suggested that including this model of the internal pressure control system within the design of an external controller would achieve better arterial pressure control and faster response than previous systems.     

Non-linear control of continuous bioreactors

Control of bioreactors has achieved importance in the recent years. This may be due to the fact that they are difficult to control which may be attributed to its nonlinear dynamic behavior. The model parameters of the bioreactor also vary in an unpredictable manner. The complexity of the biochemical processes inhibits the accurate modeling and also the lack of suitable sensors make the process state difficult to characterize. Considerable emphasis has been placed on the control of fed-batch fermentors because of their prevalence in industries. However, when production of biomass is to be optimized, continuous operation is desirable. Several procedures are available for the nonlinear control of processes, viz., differential geometric approach, internal model control approach, reference synthesis technique, predictive control design, etc., but the major disadvantage of these approaches is the computational time required to perform the prediction optimization. In this study, a nonlinear controller based on a polynomial discrete time model (NARMAX) is evaluated for its performance on a fermentor. It can be shown that a nonlinear self-tuning controller based on NARMAX model can be extended to the control of fermentors. The response is smooth for both load and setpoint changes even when process parameters are assumed to be zero and uncertainty in parameters are present and in the presence of controller constraints. The control action can be made more or less robust by changing the design parameters appropriately. Therefore, nonlinear self-tuning controller is suitable for control of industrial processes.     

A neurofuzzy inference system based on biomechanical features for the evaluation of the effects of physical training

The current study aimed to evaluate physical training effects. For this purpose, a classifier was implemented by taking into account biomechanical features selected from force-plate measurements and a neurofuzzy algorithm for data management and relevant decision-making. Measurements included two sets of sit-to-stand (STS) trials involving two homogeneous groups, experimental and control, of elders. They were carried out before and after a 12-week heavy resistance strength-training program undergone by the experimental group. Pre- and post-training differences were analysed, and percentages of membership to "trained" and "untrained" fuzzy sets calculated. The method was shown to be appropriate for detecting significant training-related changes. Detection accuracy was higher than 87%. Slightly weaker results were obtained using a neural approach, suggesting the need for a larger sample size. In conclusion, the use of a set of biomechanical features and of a neurofuzzy algorithm allowed to propose a global score for evaluating the effectiveness of a specific training program.     

MODEM: a multi-agent hierarchical structure to model the human motor control system

In this study, based on behavioral and neurophysiological facts, a new hierarchical multi-agent architecture is proposed to model the human motor control system. Performance of the proposed structure is investigated by simulating the control of sit to stand movement. To develop the model, concepts of mixture of experts, modular structure, and some aspects of equilibrium point hypothesis were brought together. We have called this architecture MODularized Experts Model (MODEM). Human motor system is modeled at the joint torque level and the role of the muscles has been embedded in the function of the joint compliance characteristics. The input to the motor system, i.e., the central command, is the reciprocal command. At the lower level, there are several experts to generate the central command to control the task according to the details of the movement. The number of experts depends on the task to be performed. At the higher level, a “gate selector” block selects the suitable subordinate expert considering the context of the task. Each expert consists of a main controller and a predictor as well as several auxiliary modules. The main controller of an expert learns to control the performance of a given task by generating appropriate central commands under given conditions and/or constraints. The auxiliary modules of this expert learn to scrutinize the generated central command by the main controller. Auxiliary modules increase their intervention to correct the central command if the movement error is increased due to an external disturbance. Each auxiliary module acts autonomously and can be interpreted as an agent. Each agent is responsible for one joint and, therefore, the number of the agents of each expert is equal to the number of joints. Our results indicate that this architecture is robust against external disturbances, signal-dependent noise in sensory information, and changes in the environment. We also discuss the neurophysiological and behavioral basis of the proposed model (MODEM).     

Interaction of epithelium with mesenchyme affects global features of lung architecture: a computer model of development.

Lung airway morphogenesis is simulated in a simplified diffusing environment that simulates the mesenchyme to explore the role of morphogens in airway architecture development. Simple rules govern local branching morphogenesis. Morphogen gradients are modeled by four pairs of sources and their diffusion through the mesenchyme. Sensitivity to lobar architecture and mesenchymal morphogen are explored. Even if the model accurately represents observed patterns of local development, it could not produce realistic global patterns of lung architecture if interaction with its environment was not taken into account, implying that reciprocal interaction between airway growth and morphogens in the mesenchyme plays a critical role in producing realistic global features of lung architecture.     

The Future of Nematology: Integration of New and Improved Management Strategies

The potential for managing plant-parasitic nenlatodes by combining two or more control strategies in an integrated program is examined. Advantages of this approach include the use of partially effective strategies and protection of highly effective ones vulnerable from nematode adaptation or environmental risk. Strategies can be combined sequentially from season to season or applied simultaneously. Programs that have several strategies available but that are limited in the true integration of control components are used as examples of current management procedures and the potential for their improvement. These include potato cyst nematodes in northern Europe, soybean cyst nematode in North Carolina, and root-knot nematodes on vegetable and field crops in California. A simplified model of the impact of component strategies on the nematode damage function indicates the potential for combining control measures with different efficacies to give acceptable nematode population reduction and crop protection. The likelihood for additive, synergistic, or antagonistic effects from combining strategies is considered with respect to the biological target and component compatibility.     

The design of controllers for batch bioreactors

The implementation of control algorithms to batch bioreactors is often complicated by variations in process dynamics that occur during the course of fermentation. Such a wide operating range often renders the performance of fixed gain proportional-integral-differential (PID) controllers unsatisfactory. In this work, detailed studies on the control of batch fermentations are per formed. Two simple controller designs are presented with the intent to compensate for changing process dynamics. One design incorporates the concepts of static feedforward-feedback control. While this technique produces tighter control than feedback alone, it is not as successful as a controller based on gain scheduling. The gain-scheduling controller, a subclass of adaptive controllers, uses the oxygen uptake rate as an auxiliary variable to fine-tune the PID controller parameters. The control of oxygen tension in the bioreactor is used as a vehicle to convey the proposed ideas, analyses, and results. Simulation experiments indicate significant improvement in controller performance can be achieved by both of the proposed approaches even in the presence of measurement noise.     

The study of neural network-based controller for controlling dissolved oxygen concentration in a sequencing batch reactor

The design and development of the neural network (NN)-based controller performance for the activated sludge process in sequencing batch reactor (SBR) is presented in this paper. Here we give a comparative study of various neural network (NN)-based controllers such as the direct inverse control, internal model control (IMC) and hybrid NN control strategies to maintain the dissolved oxygen (DO) level of an activated sludge system by manipulating the air flow rate. The NN inverse model-based controller with the model-based scheme represents the controller, which relies solely upon the simple NN inverse model. In the IMC, both the forward and inverse models are used directly as elements within the feedback loop. The hybrid NN control consists of a basic NN controller in parallel with a proportional integral (PI) controller. Various simulation tests involving multiple set-point changes, disturbances rejection and noise effects were performed to review the performances of these various controllers. From the results it can be seen that hybrid controller gives the best results in tracking set-point changes under disturbances and noise effects.     

A model for control of breathing in mammals: coupling neural dynamics to peripheral gas exchange and transport

A new model for aspects of the control of respiration in mammals has been developed. The model integrates a reduced representation of the brainstem respiratory neural controller together with peripheral gas exchange and transport mechanisms. The neural controller consists of two components. One component represents the inspiratory oscillator in the pre-Bötzinger complex (pre-BötC) incorporating biophysical mechanisms for rhythm generation. The other component represents the ventral respiratory group (VRG), which is driven by the pre-BötC for generation of inspiratory (pre)motor output. The neural model was coupled to simplified models of the lungs incorporating oxygen and carbon dioxide transport. The simplified representation of the brainstem neural circuitry has regulation of both frequency and amplitude of respiration and is done in response to partial pressures of oxygen and carbon dioxide in the blood using proportional (P) and proportional plus integral (PI) controllers. We have studied the coupled system under open and closed loop control. We show that two breathing regimes can exist in the model. In one regime an increase in the inspiratory frequency is accompanied by an increase in amplitude. In the second regime an increase in frequency is accompanied by a decrease in amplitude. The dynamic response of the model to changes in the concentration of inspired O₂ or inspired CO₂ was compared qualitatively with experimental data reported in the physiological literature. We show that the dynamic response with a PI-controller fits the experimental data better but suggests that when high levels of CO₂ are inspired the respiratory system cannot reach steady state. Our model also predicts that there could be two possible mechanisms for apnea appearance when 100% O₂ is inspired following a period of 5% inspired O₂. This paper represents a novel attempt to link neural control and gas transport mechanisms, highlights important issues in amplitude and frequency control and sets the stage for more complete neurophysiological control models.