Abstract: The availability of inexpensive, yet competent hardware allows for increased level of automation and self-optimization in the context of Industry 4.0. However, such agents require high quality information about their surroundings along with a robust strategy for collision avoidance, as they may cause expensive damage to equipment or other agents otherwise. Manually defining a strategy to cover all possibilities is both time-consuming and counter-productive given the capabilities of modern hardware. This paper explores the idea of a model-free self-optimizing obstacle avoidance strategy for multiple autonomous agents in a simulated dynamic environment using the Q-learning algorithm.
Abstract: Two-wheel inverted pendulum robot (TWIPR) is designed with two-hub DC motors for human riding and motion control evaluation. In order to measure the tilt angle and angular velocity of the inverted pendulum robot, accelerometer and gyroscope sensors are chosen. The mobile robot’s moving position and velocity were estimated based on DC motor built in hall sensors. The control kernel of this electric mobile robot is designed with embedded Arduino Nano microprocessor. A handle bar was designed to work as steering mechanism. The intelligent model-free fuzzy sliding mode control (FSMC) was employed as the main control algorithm for this mobile robot motion monitoring with different control purpose adjustment. The intelligent controllers were designed for balance control, and moving speed control purposes of this robot under different operation conditions and the control performance were evaluated based on experimental results.
Abstract: This paper presents an approach for optimal cyber security decisions to protect instances of a federated Internet of Things (IoT) platform in the cloud. The presented solution implements the repeated Stackelberg Security Game (SSG) and a model called Stochastic Human behaviour model with AttRactiveness and Probability weighting (SHARP). SHARP employs the Subjective Utility Quantal Response (SUQR) for formulating a subjective utility function, which is based on the evaluations of alternative solutions during decision-making. We augment the repeated SSG (including SHARP and SUQR) with a reinforced learning algorithm called Naïve Q-Learning. Naïve Q-Learning belongs to the category of active and model-free Machine Learning (ML) techniques in which the agent (either the defender or the attacker) attempts to find an optimal security solution. In this way, we combine GT and ML algorithms for discovering optimal cyber security policies. The proposed security optimization components will be validated in a collaborative cloud platform that is based on the Industrial Internet Reference Architecture (IIRA) and its recently published security model.
Abstract: A two wheel inverted pendulum (TWIP) vehicle is built with two hub DC motors for motion control evaluation. Arduino Nano micro-processor is chosen as the control kernel for this electric test plant. Accelerometer and gyroscope sensors are built in to measure the tilt angle and angular velocity of the inverted pendulum vehicle. Since the TWIP has significantly hub motor dead zone and nonlinear system dynamics characteristics, the vehicle system is difficult to control by traditional model based controller. The intelligent model-free fuzzy sliding mode controller (FSMC) was employed as the main control algorithm. Then, intelligent controllers are designed for TWIP balance control, and two wheels synchronization control purposes.
Abstract: Modeling the behavior of the dialogue management in
the design of a spoken dialogue system using statistical methodologies
is currently a growing research area. This paper presents a work
on developing an adaptive learning approach to optimize dialogue
strategy. At the core of our system is a method formalizing dialogue
management as a sequential decision making under uncertainty whose
underlying probabilistic structure has a Markov Chain. Researchers
have mostly focused on model-free algorithms for automating the
design of dialogue management using machine learning techniques
such as reinforcement learning. But in model-free algorithms there
exist a dilemma in engaging the type of exploration versus exploitation.
Hence we present a model-based online policy learning
algorithm using interconnected learning automata for optimizing
dialogue strategy. The proposed algorithm is capable of deriving
an optimal policy that prescribes what action should be taken in
various states of conversation so as to maximize the expected total
reward to attain the goal and incorporates good exploration and
exploitation in its updates to improve the naturalness of humancomputer
interaction. We test the proposed approach using the most
sophisticated evaluation framework PARADISE for accessing to the
railway information system.
Abstract: A model-free robust control (MFRC) approach is proposed for position control of robot manipulators in the state space. The control approach is verified analytically to be robust subject to uncertainties including external disturbances, unmodeled dynamics, and parametric uncertainties. There is a high flexibility to work on different systems including actuators by the use of the proposed control approach. The proposed control approach can guarantee the robustness of control system. A PUMA 560 robot driven by geared permanent magnet dc motors is simulated. The simulation results show a satisfactory performance for control system under technical specifications. KeywordsModel-free, robust control, position control, PUMA 560.
Abstract: A novel low-cost impedance control structure is
proposed for monitoring the contact force between end-effector and
environment without installing an expensive force/torque sensor.
Theoretically, the end-effector contact force can be estimated from the
superposition of each joint control torque. There have a nonlinear
matrix mapping function between each joint motor control input and
end-effector actuating force/torques vector. This new force control
structure can be implemented based on this estimated mapping matrix.
First, the robot end-effector is manipulated to specified positions, then
the force controller is actuated based on the hall sensor current
feedback of each joint motor. The model-free fuzzy sliding mode
control (FSMC) strategy is employed to design the position and force
controllers, respectively. All the hardware circuits and software
control programs are designed on an Altera Nios II embedded
development kit to constitute an embedded system structure for a
retrofitted Mitsubishi 5 DOF robot. Experimental results show that PI
and FSMC force control algorithms can achieve reasonable contact
force monitoring objective based on this hardware control structure.
Abstract: Since the driving speed and control accuracy of
commercial optical disk are increasing significantly, it needs an
efficient controller to monitor the track seeking and following
operations of the servo system for achieving the desired data
extracting response. The nonlinear behaviors of the actuator and servo
system of the optical disk drive will influence the laser spot
positioning. Here, the model-free fuzzy control scheme is employed to
design the track seeking servo controller for a d.c. motor driving
optical disk drive system. In addition, the sliding model control
strategy is introduced into the fuzzy control structure to construct a
1-D adaptive fuzzy rule intelligent controller for simplifying the
implementation problem and improving the control performance. The
experimental results show that the steady state error of the track
seeking by using this fuzzy controller can maintain within the track
width (1.6 μm ). It can be used in the track seeking and track
following servo control operations.
Abstract: A self tuning PID control strategy using reinforcement
learning is proposed in this paper to deal with the control of wind
energy conversion systems (WECS). Actor-Critic learning is used to
tune PID parameters in an adaptive way by taking advantage of the
model-free and on-line learning properties of reinforcement learning
effectively. In order to reduce the demand of storage space and to
improve the learning efficiency, a single RBF neural network is used
to approximate the policy function of Actor and the value function of
Critic simultaneously. The inputs of RBF network are the system
error, as well as the first and the second-order differences of error.
The Actor can realize the mapping from the system state to PID
parameters, while the Critic evaluates the outputs of the Actor and
produces TD error. Based on TD error performance index and
gradient descent method, the updating rules of RBF kernel function
and network weights were given. Simulation results show that the
proposed controller is efficient for WECS and it is perfectly
adaptable and strongly robust, which is better than that of a
conventional PID controller.
Abstract: The majority of existing predictors for time series are
model-dependent and therefore require some prior knowledge for the
identification of complex systems, usually involving system
identification, extensive training, or online adaptation in the case of
time-varying systems. Additionally, since a time series is usually
generated by complex processes such as the stock market or other
chaotic systems, identification, modeling or the online updating of
parameters can be problematic. In this paper a model-free predictor
(MFP) for a time series produced by an unknown nonlinear system or
process is derived using tracking theory. An identical derivation of the
MFP using the property of the Newton form of the interpolating
polynomial is also presented. The MFP is able to accurately predict
future values of a time series, is stable, has few tuning parameters and
is desirable for engineering applications due to its simplicity, fast
prediction speed and extremely low computational load. The
performance of the proposed MFP is demonstrated using the
prediction of the Dow Jones Industrial Average stock index.