Energy-aware data collection is of paramount importance for robotic and wireless sensor networks. Although static sink-aided cluster-based protocols provide energy-efficient solutions, unmanned aerial vehicle (UAV)-aided approaches can be considered as better alternatives to reduce energy consumption while data acquisition compared with static sinks. Most of the existing UAV-driven solutions have not considered a limit on the battery capacity of the UAV, which needs to be considered in a practical manner. This article investigates energy-aware data collection in robot network clusters. In each cluster, a cluster head (CH) robot allocates one collaborative task to each cluster member (CM) robot and collects data from CMs whereas a UAV collects data from CH robots by visiting a subset of them due to its battery limitation. To complement the state-of-the-art, UAV decision for visiting the subset of CHs is constrained to multiple factors including residual battery capacity, as well as locations and data qualities of all CH robots. Nonvisited CH robots use CH robots as relay nodes for data forwarding. Following upon this, by considering the problem under data hopping constraints, this article also presents a sensitivity analysis with respect to data hopping constraints. Simulations show that the proposed policy achieves zero total joint cost whereas the state-of-the-art approaches result in significantly high total joint costs. Furthermore, the proposed policy reduces the total joint cost by up to 50% with respect to the conventional approaches.

This paper focuses on reshaping a previously detected human intention into a desired one, using contextual motions of mobile robots, which are in our applications, autonomous mobile 2-steps stairs and a chair. Our system first estimates the current intention based on human heading and trajectory depicted as orientation and location. Our previous reshaping applications have shown that the current human intention has to be deviated towards the new desired one in phases. In our novel approach, Elastic network generates way points of trajectories each of which acts as transient trajectories directed towards the desired intention's location. Our methodology aims at generating an “intention trajectory” towards the final goal. The initial way points possess destabilizing effects on the obstinance of the person intention making the “robot gain the curiosity and the trust of the person”. Each way point generated by the elastic network is executed by moves of an adequate robot (here mobile 2-steps or chair) in adequate directions (towards coffee table, PC, TV, library). After each robot moves, the resulting human intention is estimated and compared to the desired goal in the intention space. Intention trajectories are searched in two modes: the “confident mode” and the “suspicious mode” which are defining human body-mood detected relying on proxemics. This paper analyzes our novel approach of planning trajectories via elastic networks based on these two modes.

In this paper we design and implement a 2D coupled mobility snake/caterpillar vehicle for terrains with variable friction. A control hierarchy made of fuzzy logic controllers equipped with fuzzy actor critic learning (FACL) enables the robot to adapt itself to variable friction conditions of unknown, unstructured environments as well as the need for task changes in those environments such pushing aside a debris for its removal from the navigation site, during locomotion.

The paper presents a fluid dynamics based framework for control of emergent behaviors of robot swarms that are modeled as fluids. A distributed low-level control mechanism is developed based on Smoothed Particle Hydrodynamics (SPH) and it is coupled with a high-level control layer that is responsible for the control of fluid parameters to generate desired behaviors from the swarming characteristics of the robots. It is shown by simulations that using the same low-level SPH model, different swarming behaviors can emerge from the local interactions of robots according to the settings of the fluid parameters that are controlled by the high-level control layer.

This work focuses on the design and implementation of a behavior based, layered, cooperative control of robots in a sensor network, hybrid in combining subsumption and motor schema layers. Behaviors at different level of resolution are not only coordinated by the motor schema layers but also through units called "evaluators" which fuses state and activity of behaviors according to priorities and cross-couplings in runtime. Concurrent, sequential, asynchronous multi-robot task allocation is based on "market based auction algorithm" with a novel efficient non-linear time varying fitness calculation enhancing fairness and fault tolerance. A 3D simulation environment is developed to test the proposed architecture for fairness, fault tolerance, mission accomplishment rates, and coverage

In this paper, we develop a distributed intelligent fuzzy learning controller for serpentine locomotion while doing lasso-type grasping in 3D environments. The novelty of our approach is not only in the control modularity but on a lasso type grasping having as a moving base the remaining snake links undergoing serpentine gait. Dynamical constraints due to environmental conditions are considered online by a genetic algorithm based evaluator, which evaluates the feasibility of behaviors suggested by the high level controller and modify them. Our approach is demonstrated on the simulation of our modular 12 link snake robot

We propose a novel approach inspired by compressible fluids as a distributed and scalable solution to the deployment problem of mobile sensor networks in unknown environments. Our approach is based on fluid dynamics through which we model the sensor network as a fluid body and each sensor node as a fluid element. Originating from local neighborhood interactions, the principles of fluid flow are adapted to the deployment of the sensor network. Mimicking the diffusive and self-spreading behavior of compressible fluids in a mobile sensor network, we achieve the desirable properties of effective coverage, scalability, and distributed self-deployment. An adaptive deployment scheme is also devised for urban disaster environments where the number of nodes to be deployed may not be determined a priori. Simulation results show that our approach can guarantee a desired level coverability

In this paper, a distributed intelligent fuzzy learning controller architecture is developed for a snake robot to avoid obstacles while reaching a target, in a dynamic environment. Moreover, in order to use the robot in object carrying search and rescue (SAR) applications, a grasping scheme of desired objects is added to the abilities of the controller. Behavior-based approach is also incorporated by "target reaching", "link following", "obstacle avoidance" and "object grasping" behaviors. Each link of the robot is separately controlled and able to select a behavior at any time. Behaviors are all realized as fuzzy actor critic learning (FACL) controllers (Jouffe, 1998). In order to move the robot in between different configurations and take dynamics into account, an online optimization technique based on genetic algorithms is employed.

In this paper a gait synthesizer for a hexapod robot working in search and rescue operations is designed. The gait synthesizer is composed of three modules responsible for selecting a new gait, evaluating tide current one, and modifying the recommended gait according to the internal reinforcements of past time steps. A fuzzy logic controller is implemented in selecting the new gaits. Different hexapod wave gaits are encoded within its fuzzy rules to model the gaits in the rule base of the controller. Simulation results demonstrate the potential of the gait synthesizer for search and rescue operations such as its adaptability to the uneven terrain by shaping gaits, getting out of an entrapment of a leg, modifying gaits when some legs are used as manipulators.

Our aim in this work is to generate an intelligent interface for profiling and evaluating each student remotely accessing a robot assisted lab setup. The interface provides the student with the "hands on" experience by necessary visual feedback, giving this user as much freedom as possible to control the experiment. Besides, the system evaluates the user's performance, adapts the context to the level of user's acquired knowledge and skill, and thus intelligently coaches the user to successfully do the experiment. The system's ability is demonstrated on an example and the behavior of each module in the proposed system is discussed. The major contribution of this paper is taking the student behavior into consideration while providing online robotics experimentation, evaluating the user's performance and coaching him/her towards the successful achievement of the tasks.

We propose in this paper a method to enhance the performance of a coupled global positioning/inertial navigation system (GPS/INS) for both air and land navigation applications during the GPS signal loss. Our method is based on using an artificial neural network (ANN) to intelligently aid the GPS/INS coupled navigation system in the absence of GPS signals. The proposed enhanced GPS/INS is used in the very dynamic environment of a tour of an autonomous van on the METU campus and we provide the results here. Our GPS/INS+ANN system performance is thus demonstrated with the land trials.

Develops a new gait synthesizer for a snake-like robot in an unstructured changing environment similar to earthquake rubbles. The gait synthesizer, inspired from the GARIC Architecture, is implemented on a simulated navigation of a snake robot. We consider 6 prototypical gaits classified into vertical undulation, lateral undulation, heading changing right/left, flapping right/left. The gait synthesizer is composed of a gait selection network that decides on a combination of prototypical gaits based on a fuzzy controller which is further tuned by a gait evaluation module based on an ANN that learns from prior performances of robot locomotion and acts as an arbiter unit on gait performance. The third module is a stochastic gait modifier, which takes feedback information from the dynamic robotic system and changes the gait based on reinforcement learning.

This paper presents a logical sensor to be used in the intelligent perception for the motion planning of a mechanical snake searching for rescue in an unstructured environment of collapsed buildings after a natural disaster. The multi-link robot is equipped with ultrasound sensors and a thermal sensor, which detects body heat and isolates any survivor. In this study, the objective is to develop a logical sensing and an intelligent perception model that learns to identify the most suitable zone for control by minimizing the control unpredictability due to uncertainty. This model is based on a radial basis function network trained with dynamic decay adjustment algorithm that classifies the control region into positive regions. Uncertainty irregularities in the control region is modeled using the statistical rough set theory based on beliefs assigned to control cells in a cellular space. Logical sensors are then fused together to construct a wide-angle logical sensor unit in the control zone. Simulation results are given, in order to illustrate the efficiency of the approach developed for an unstructured environment.

We introduce an online transition detection method of degradation of control predictability to robotic systems which can behave chaotically depending on the variations of the control parameter. This method allows us to control such systems at the boundary of the chaotic region until the onset of degradation thus allowing us to use the whole richness of information for control. In our approach, a safe control parameter interval is chosen initially where the system is certainly nonchaotic and the system behavior is initiated. The maximum Lyapunov exponents corresponding to each parameter value are estimated online. The parameter interval is then widened according to the estimated exponents until the maximum exponent approaches a small negative value. At this value, the estimation is ended and the controller begins to work in this maximum possible controller parameter interval. This method allows the designer to have a more efficient controller and a more agile system. This approach is applied to a robot arm under dynamic load.

This paper presents the multisensor-based online path planning of a serpentine robot in the unstructured changing environment of earthquake rubble during search for survivors. The robot has 6 links and 16 controllable degrees of freedom and is equipped with an ultrasound sensor and a camera on its head and infrared sensors along the links of the robot for obstacle avoidance. The serpentine robot locomotion is achieved as a combination of basic snakelike and wormlike gaits which are that of lateral undulation, rectilinear motion, flapping motion (right, left), and of torsional rotation (which has the pivot point at the center of the robot). The path planning algorithm is adopted from distance transform based exploratory path planning and modified to our serpentine robotic configuration. The adopted method is used for calculating the cost of the possible next configuration states so that the snake/worm like robot selects its next gait accordingly. Environmental data is obtained from the multiple sensors located along the body of the robot. The sensors are located in the following configuration: one ultrasound sensor for detecting the obstacles in front of the robot and constructing the map for free space, and 12 infrared sensors (6 pairs) which are located on the left and right of the joints of the robot. Simulation results are presented in the paper as an illustration of the efficiency of the method in complex unstructured environments.

We introduce in this paper a new fractal/rough set modeling approach to the domains of attraction of nonlinear systems obtained by cell mapping. The state space is partitioned into cells and the stability regions found using cell to cell mapping. Our new approach gives a fractal rough set identity to the domains of attraction where cells are identified according to their fractal dimension as fully stable, possibly stable and unstable. There the stability domain is a rough set where fully stable cells determine the lower approximation of the domain, and possibly stable cells its rough boundary. Consequently, the totality of these cells forms an upper approximation to the rough stability domain. The boundary of this domain which is a rough set of cells having a fractal dimension as an attribute of roughness is smoothed, minimising the inherent stability uncertainty of the region, using a reinforcement learning technique which takes into account the stability history of each fractal/rough cell. This new approach intended to reinforce the performance of a controller under stability uncertainty is applied for illustrative purposes to a two-axis robot arm under dynamic load.

This paper introduces a methodology developed for the self-localization of a holon, within the reconfiguration task space of its colony under a highly decentralized decision-making and colony reconfiguration control. Holons act as intelligent agents of a colony in this decentralized structure. We term such an n-holon colony where intelligent agents can self-localize in reconfiguration task space without their tasks within a configuration being assigned by a central controller, as an n-holony. Our approach begins the self localization in task domain of a holon by proposing a way of extracting the invariants of the global state of the holony configuration which will then be used to infer the localization and action estimation by a holon using local knowledge. The holon communicates with its neighbours in order to modify its local knowledge towards a global one. This is the basic inference structure of a holony that is novel when compared to that of a colony of holons.

This paper uses a holonic hand model introduced in our early works (1998), which is composed of 2-stems-of-links-1-joint nonconvex biped planar holons working in colony in order to perform dexterous grasps. This paper focuses on the reconfiguration control of the colony that we structure as a sequence of planar grasp operations defining a holonic regrasp, which uses a graph theoretic formalization of configuration changes that occur in this holonic hand during this regrasping phase.

This paper introduces a methodology developed for the self-organization of a colony of holons in its task space where holons act as intelligent agents in this decentralized robotic structure such that they self localize themselves in the reconfiguration task space of their colony and identify their individual role in the colony re-organization. We term such an n-holon colony where intelligent agents can self-localize in the reconfiguration task space without their individual tasks within the colony configuration being assigned by a central controller, an n-holony. The proposed method is based on a cascaded computation scheme for the self-reconfiguration of the n-holony: self-localization followed by self-organization. The holon communicates with its neighbours in order to acquire knowledge about the colony configuration within a neighbourhood of radius d based on configuration invariants namely the eigenstructures. Self-organization uses this knowledge to take the next action for achieving the desired holony configuration.

Many algorithms have been proposed in order to train radial basis function (RBF) networks. Resilient Backpropagation (RPROP) with a weight decay term is proposed to train RBF networks and used to differentiate surfaces of 3D objects in range images and to classify eight different machine learning data sets for classification purposes. We show the advantages of resilient backpropagation for the RBF network structure within this classification context. The network structure is a combination of supervised and unsupervised learning layers. Experimental results show that a radial basis function network trained with resilient backpropagation can be successfully applied to differentiate surfaces of 3D objects in range images as well as to classification of machine learning problems.

We develop a 2-way adaptive fuzzy control system that makes use of the intuitionistic fuzzy sets for modeling expert knowledge bearing uncertainty. Adaptive fuzzy control systems are fuzzy logic systems whose rule parameters are automatically adjusted through training. In our system, all supports to propositions have interval valued distributions with necessity at the lower bound and possibility at the upper. Uncertainty in expert knowledge determines the width of the interval. Our first level training tunes rule parameters with necessity function values, while the second level training re-adjusts these parameters so as to minimize uncertainty based on possibility function values. The intuitionistic 2-way adaptive fuzzy controller is found to have a better performance due to the richness of information in interval valued supports with nonantagonistic bounds opposed to antagonistic ones in ordinary fuzzy sets.

A primary characteristic of dexterity in the manipulation of objects by multifingered hands in uncertain environments is the ability to recover from failing grasps through a regrasp procedure. Our focus in this paper is on regrasping control, posterior to the grasp phase, by generating an optimal sequence of hand preshapes, close in terms of manipulability and stability while not losing contact with the object. In our approach we treat the preshaping of a hand and the manipulation of an object with the same hand as a continuous and inseparable process, formulate preshaping manipulability and stability criteria using the vortex theory, and model the problem of regrasping as an optimal search using genetic algorithms toward a predetermined final preshape initiated by the kinematic constraints of a prior preshaping that has optimally contacted an object.

This paper investigates the holonic modeling of multifingered robotic hands where 2-stem-of-links+1-joint holons having basic kinematic properties are considered, working in colony in order to perform dexterous grasps. Those holons are the simplest physical elements, which have the capabilities of independent actuation, and of cooperation with other holons in achieving global behavior for the whole structure. Our work focuses on the concept of self-reconfigurability of the modular colony structure so as the holons reconfigure themselves to a pre-described structure in the absence of a human in the loop. The reconfiguration control of the dynamical holonic hand is a regrasp control and is developed based on hybrid control since our system exhibits both discrete and continuous behaviors.

Proposes an escape methodology to the local minima problem of self-organizing feature maps generated in the overlapping regions which are equidistant to the corresponding winners. Two new versions of the self-organizing feature map are derived equipped with such a methodology. The first approach introduces an excitation term, which increases the convergence speed and efficiency of the algorithm while increasing the probability of escaping from local minima. In the second approach we associate a learning set which specifies attractive and repulsive fields of output neurons. Results indicate that accuracy percentile of the new methods are higher than the original algorithm while they have the ability to escape from local minima.

This paper focuses on the development and implementation of a Genetically Tuned Fuzzy Scheduler (GTFS) for heterogeneous FMS under uncertainty. The scheduling system takes input from a table and creates an optimum master schedule. The GTFS uses fuzzy rulebase and inferencing where fuzzy sets are generated by a genetic algorithm to tune the optimization. The fuzzy optimization is based on time criticality in deadline and machine need, taking into account machine availability, uniformity, process time and selectability.

The objective of the paper is to generate the necessary subsystems for the intelligent guidance of an airborne vehicle locking onto a swift target. This paper focuses on the generation of an intelligent tracker module equipped with a wavelet based neural network that learns predictions from past experience. The perception of actual target manoeuvre and prediction of its future states are achieved in this work by "projecting" actual observations into decision spaces of local fuzzy predictions based on independent prototypical trajectory types: linear, parabolic and square root type trajectory. Decentralized tracking decisions are thus generated which are further evaluated by learning prediction module and are fused before being sent to the guidance module. The tracker is tested for 3 dimensional target tracking problem.

We introduce a hybrid intelligent karyotyping system based on two different types of artificial neural networks (ANNs) and chromosome's features obtained by digital image processing techniques. A microscope equipped with a CCTV camera and a microcomputer including a frame grabber are the basic components of our hardware set-up. The inputs to the ANN structure are obtained directly from digital chromosome images by using two recently developed object detection and object skeletonizing algorithms. Moreover, the band patterns of chromosomes are represented by applying wavelet transform techniques on the gray level profiles of chromosomes. The network parameters are determined by using the results of many training and testing experiments in order to reach an optimal state from the classification performance point of view.

The authors' work focuses on developing methods of determining the optimal preshape of a robot hand for a particular stability and manipulability degree based on kinematic considerations. To this end, for a given task, the authors define stability and manipulability criteria for a robot hand based vortex theory. The effects of these criteria and manipulation to the hand preshape are examined by generating optimal fingertip landing paths on an object. This object is a perceived 2D contour in the robot workspace and the optimal landing trajectories are determined according to required manipulability and stability at the preshaping phase to be imparted to the object at contact. An illustrative example of optimal preshaping is provided at the end of the paper and the effects of the vortex based stability and manipulability are discussed in the results.>

Proposed is a novel model of belief functions based on fractal theory. The model is first justified in qualitative, intuitive terms, then formally defined. Also, the application of the model to the design of an evidential classifier is described. The proposed classification scheme is illustrated by a simple example dealing with robot sensing. The approach followed is motivated by applications to the design of intelligent systems, such as sensor-based dexterous manipulators, that must operate in unstructured, highly uncertain environments. Sensory data are assumed to be (1) incomplete and (2) gathered at multiple levels of resolution.>

Dexterous grasps for multifingered robot hands are planned. A real-time, joint space finger path planning algorithm is derived for the enclosure phase of grasping motion. The algorithm minimizes the impact momentum of the hand. It uses a preshape Jacobian matrix to map task-level hand preshape requirements into kinematic constraints. A master-slave scheme avoids inter-finger collisions and reduces the dimensionality of the planning problem.>