Data-Driven Inverse Kinematics (DDIK) solvers emerged as promising Inverse Kinematics (IK) methods for reliably approximating the IK of robotic manipulators. However, these solvers remain heavily robot-dependent, where for each robot of interest, a network needs to be trained in an one-solver-one-robot framework. In this paper, we build on our previous work on Learning-By-Example (LBE) for DDIK, and introduce an one-solver-many-robots framework; where a single neural network is used to predict the IK of multiple robots – mainly with 6 and 7 Degrees of Freedom (DoF). In our LBE approach, the neural network input includes an example of joint-pose tuple (e.g. any previous joint and corresponding pose tuple in the path) along with the queried pose as the same network outputs the desired robot joint configuration. Here, we investigate five network architectures: a Plain Multilayer Perceptron (MLP), a Residual-Based MLP (RMLP), a Densely Connected MLP (DMLP), and two transformers inspired by Generative Pre-Trained Transformer (GPT) and tested them using 3 diverse datasets with 20 real-world robotic arms with 6 and 7 DoF. Our experimental results demonstrate that a single lightweight, LBE-based DDIK solver can reliably predict the IK for multiple and hitherto unseen robots, within each of the 6 or 7DoF family as well as across both 6 and 7DoF robot families with position errors below 1mm and orientation errors below 1deg. Additionally, we compare all proposed LBE-DDIK solvers with three established numerical IK solvers: Selectively Damped Least-Squares (SD), Singular Value Filtering (SVF), and Mixed Inverse (MX) and observe that our LBE-DDIK solvers achieve comparable accuracy, with the advantage of being a one-solver-many-robots framework.

Local Binary Pattern (LBP) and its variants have considerable success in a wide range of computer vision and pattern recognition applications, especially in tasks related to texture classification. However, the LBP method is sensitive to noise, scale variations and unable to capture macro-structure information. We propose a novel texture classification descriptor called Scale Adaptive Robust LBP (SARLBP) that enhances macro-level descriptive information by incorporating significantly larger scales, and a novel encoding scheme, which is designed to overcome the limitations of traditional LBP schemes. SARLBP method dynamically determines a single optimal scale for each radial direction from multiple scales based on the local area's characteristics. Subsequently, this descriptor extracts four distinct patterns derived from regional image medians of center pixel, radially-optimized neighbor pixels, optimized fixed scale-based pixels, and radial-difference-based pixels. This method adeptly captures texture information at both micro and macro scales by employing scale adaptation based on the distinctive attributes of the local region. As a result, it provides a comprehensive and robust representation of the texture images. Extensive experimentation was conducted on four publicly available texture databases (ALOT, CUReT, UMD, and Kylberg), considering both the presence and absence of two distinct types of interference (Gaussian noise and Salt-and-Pepper noise). The results reveal that our SARLBP method achieves significantly better performance than other state-of-the-art LPB variants with a fixed smaller feature dimension.

Outlier interpretation is essential in various data processing fields within the context of outlier detection. Under-standing the specific attributes that make an outlier distinct is particularly valuable in many application areas, especially where, due to class imbalance, it can greatly enhance analytic insights into unusual occurrences or trends. To address this need, we introduce a Regularized Auto-Encoder with a Genetic Algorithm (RAE-GA) approach for outlying and characterizing subspaces: i.e. a novel approach for identifying outliers and pinpointing their underlying subspaces. This method leverages autoencoders with regularized latent spaces, such as in Variational Auto Encoders (VAE) and Regularized Auto Encoders (RAE), which have proven effective in outlier detection only. So, building upon this success, the proposed RAE-GA utilizes a specially crafted, computationally efficient fitness function based on RAE. This function is integrated into the genetic algorithm to explore various potential subspaces, eliminating the need for restrictive pruning or constraints. As our experimental results indicate, RAE-GA outperforms the current state-of-the-art methods in outlier interpretation.

Performing feature selection on a small number of instances with high-dimensional datasets poses a needed challenge in preventing over-fitting. To address this issue, this paper proposes a sequential transfer-learning approach combined with a multi-objective genetic algorithm (STMO-GA) for feature selection. Firstly, for the multi-objective component of our method, we employ a Non-dominated Sorting Genetic Algorithm (NSGA-II) to generate a Pareto front. Then, features are ranked based on their number of appearances in the same Pareto front. Next, during the sequential knowledge transfer process, the ranked features are iteratively selected until a predetermined$n$number of features remains. This feature subspace is further refined by a k-fold cross-validation operation, starting from the rank-one feature, to determine the cut-off of the$n$features that will remain. Comparative evaluations against both GA-based as well as traditional feature selection methods demonstrate that the proposed method achieves superior classification accuracy, while retaining the smallest number or a comparable number of features.

Classification and segmentation using ultra-fine-grained datasets can be challenging due to the small nuances between adjacent classes. This problem can be exacerbated by the fact that variations within classes can be much larger than other variations between classes. Some approaches have resorted to attention mechanisms that focus on the source or the properties of the features that cause these minor changes in samples between or within classes. In some cases, the attention mechanism can be derived from spatial, temporal, modal, or other types of features in the dataset. Sometimes, attention can be drawn from external sources such as the shape of the object, its skeleton, contour, etc. Finally, some approaches use completely independently extracted information to guide the attention mechanism in a supervised fashion (privileged information, guided-attention, etc). In this paper, we claim that in the context of ultra-fine datasets with a small number of samples, a simple attention mechanism can improve the classification results. Moreover, the same simple attention mechanism can be employed in a backbone topology for the segmentation of the same information that would otherwise be used to guide the attention mechanism in other methods. In other words, unlike the state-of-the-art model for ultra-fine-grained classification of, for example, plant leaves datasets, which uses segmentation masks to guide its attention mechanism, our proposed network can simultaneously provide a classification label and a segmentation mask. The XGBoost algorithm was applied to the attention-modulated feature map for classification, and the Optuna hyperparameter optimization framework was used to tune XGBoost. Three state-of-the-art methods were compared against ours using three benchmark datasets, and our model, XMNet, achieved the best results for the vein segmentation task. For the classification part, our network achieved comparable performance with respect to two state-of-the-art as well as various other more traditional methods.

Inverse Kinematics (IK) is one of the most fundamental challenges in robotics. It refers to the process of determining the joint configurations required to achieve the desired position and orientation (pose) of a robot end-effector. Although numerous Data-Driven (DD) IK solvers have demonstrated encouraging results, they have not achieved the same accuracy when compared to other IK methods for complex robot configurations (e.g., numerical methods for higher Degrees of Freedom (DoF)). In this work, we propose a new Learning-by-Example method, and show that such a scheme considerably improves the IK learning results when compared to other DD learners. In our approach, the network input incorporates an example of joint-pose pair along with the query pose to predict the desired robot joint configuration. We show that the example joint-pose pair does not need to be too close to the query – i.e. example and query can be as far as 20 degrees apart in the joint configuration space. Furthermore, we investigate the utilization of residual and dense skip connections in Multilayer Perceptron for DDIK solvers and employ the resulting networks for two redundant robotic manipulators: a 7-DoF-7R commensurate robot and a 7DoF-2RP4R incommensurate robot. Our experimental results show that the resulting DDIK solver can reliably predict IK solutions with accuracy better than 1mm in position and 1deg in orientation.

As machine learning solutions become increasingly more ubiquitous in medical diagnosis, researchers are becoming equally more aware of the possibility of confounded predictions being produced by these same models. This realization derives, for example, from the observation that sample-wise crossvalidation leads to highly underestimated error predictors when compared to subject-wise cross validation. However, without a reliable approach to remove spurious, confounding factors such as age, gender, or even the type/brand of equipment used, these same machine learning solutions will be fated to produce poor results despite the metrics for error estimation employed. In this research, we propose an optimization approach, using genetic algorithms, to adapt the feature vectors in order to maximize the prediction accuracy of a given classifiers, while minimizing the correlation between the features and the potential confounding factors. Our results, when applied to the diagnostic of vocal fatigue, have shown great improvement in terms of the generalization capability of the chosen SVM classifier. The system was evaluated using subject-wise (i.e. leave-one-subject-out) cross-validation, which demonstrated the effectiveness of this new confounding removal approach.

One of the most important problems in robotics is the computation of the inverse kinematics (IK). This apparently simple task is necessary to determine how to move each joint in order to reach a desired end-effector pose in Cartesian space. However, the associated forward kinematics can be a highly nonlinear, non-bijective, and multidimensional function for which it may be difficult or even impossible to find closed-form solutions for its inverse - especially as the number of Degrees of Freedom (DoF) increases. Several approaches have been taken using non-linear approximators to solve IK problems. In this paper, we present a study on solving the inverse kinematics of multiple robotic arms using Artificial Neural Networks (ANN) and Adaptive Neuro-Fuzzy Inference Systems (ANFIS). For this study, we experimented with 4, 5, 6 and 7 DoF serial robots, with combinations of prismatic and revolute joints. Unlike other task-oriented solvers, our goal was not to predict poses based on specific trajectories (linear or otherwise), but instead to learn the entire robot workspaces. This goal better addresses real-world uses of robotic IK, where any end-effector pose should be reachable from any current pose. From the experiments conducted, we conclude that both ANN and ANFIS converged to some degree to the underlying inverse kinematics function, however approximation errors and the time and effort required to achieve those results may not justify their use vis-a-vis other methods in the literature.

Symptoms of voice disorder may range from slight hoarseness to complete loss of voice; from modest vocal effort to uncomfortable neck pain. But even minor symptoms may still impact personal and especially professional lives. While early detection and diagnosis can ameliorate that effect, to date, we are still largely missing reliable and valid data to help us better screen for voice disorders. In our previous study, we started to address this gap in research by introducing an ambulatory voice monitoring system using surface electromyography (sEMG) and a robust algorithm (HiGUSSS) for pattern recognition of vocal gestures. Here, we expand on that work by further analyzing a larger set of simulated vocal dysfunctions. Our goal is to demonstrate that such a system has the potential to recognize and detect real vocal dysfunctions from multiple individuals with high accuracy under both intra and intersubject conditions. The proposed system relies on four sEMG channels to simultaneously process various patterns of sEMG activation in the search for maladaptive laryngeal activity that may lead to voice disorders. In the results presented here, our pattern recognition algorithm detected from two to ten different classes of sEMG patterns of muscle activation with an accuracy as high as 99%, depending on the subject and the testing conditions.

This paper is an improvement of our previous work [1]. It provides a robust, fast and accurate solution for the inverse kinematics problem of generic serial manipulators - i.e. any number and any combination of revolute and prismatic joints. Here, we propose further enhancements by applying an evolutionary approach on the previous architecture and explore the effects of different parameters on the performance of the algorithm. The algorithm only requires the Denavit-Hartenberg (D-H) representation of the robot as input and no training or robot-dependent optimization function is needed. In order to handle singularities and to overcome the possibility of multiple paths in redundant robots, our approach relies on the computation of multiple (parallel) numerical estimations of the inverse Jacobian while it selects the current best path to the desired configuration of the end-effector using an evolutionary algorithm. But unlike other iterative methods, our method achieves submillimeter accuracy in 20 iterations in average. The algorithm was implemented in C/C++ using POSIX threads, and it can be easily expanded to use more threads and/or many-core GPUs. We demonstrate the high accuracy and real-time performance of our method by testing it with five different robots including a 7-DoF redundant robot. Results show that the evolutionary implementation of the algorithm is able to reduce the number of iterations compared to the previous method significantly, while also finding the solution within the specified margin of error.

Voice disorders are non-trivial when it comes to their early detection. Symptoms range from slight hoarseness to complete loss of voice, and may seriously impact personal and professional life. To date, we are still largely missing reliable data to help us better understand and screen voice pathologies. In this paper, we present an ambulatory voice monitoring system using surface electromyography (sEMG) and a robust algorithm for pattern recognition of vocal gestures. The system, which can process up to four sEMG channels simultaneously, also can store large amounts of data (up to 13 hours of continuous use) and in the future will be used to analyze on-the-fly various patterns of sEMG activation in the search for maladaptive laryngeal activity that may lead to voice disorders. In the preliminary results presented here, our pattern recognition algorithm (Hierarchical GUSSS) detected six different sEMG patterns of activation, and it achieved 90% accuracy.

In our previous work, a new method for measuring limb volume based on infrared depth sensors was presented. The system, which can be operated in the comfort of our homes, allows for the early detection of swelling associated with lymphedema - a chronic disease caused by failure in the lymphatic system. Early detection and management can significantly reduce the potential for symptoms and complications; however, many patients fail to seek medical assistance at the first sign of the disease. So, the proposed system can potentially affect the lives of nearly 500,000 people in the U.S. who suffer from lymphedema with over 2.6 million breast cancer survivors and over 230,000 new cases every year1who are at-risk for developing this disease at some point in their life. In this paper, a series of improvements made to the system is presented. The changes led to the complete automation of the process of 3D imaging the arms. The proposed technique for limb-volume measurement was compared with the water displacement and the perometry. Being an ongoing research, the results presented here are limited to 14 arms of healthy volunteers. In the future, test will include a larger number of limbs of healthy as well as cancer patients.

We propose a robust and fast solution for the inverse kinematic problem of general serial manipulators - i.e. any number and any combination of revolute and prismatic joints. The algorithm only requires the Denavit-Hartenberg (D-H) representation of the robot as input and no training or robot-dependent optimization function is needed. In order to handle singularities and to overcome the possibility of multiple paths in redundant robots, our approach relies on the computation of multiple (parallel) numerical estimations of the inverse Jacobian while it selects the current best path to the desire configuration of the end-effector. But unlike other iterative methods, our method achieves sub-millimeter accuracy in 20.48ms in average. The algorithm was implemented in C/C++ using 16 POSIX threads, and it can be easily expanded to use more threads and/or many-core GPUs. We demonstrate the high accuracy and the real-time performance of our method by testing it with five different robots, at both non-singular and singular configurations, including a 7-DoF redundant robot.

We propose a novel human-computer interface for video gaming systems. Unlike current systems that use active sensors (e.g. Kinect), in the proposed approach, the human-poses are captured by a single camera and used to control the course of the game. This eliminates the need for traditional hand-held controllers and allows for the player to be immersed into the virtual environment of the game. The success of the approach is achieve through a robust extraction of human silhouettes and by a simple but effective human-pose recognizer. The system employs an adaptive framework for background subtraction using Adaptive Local Principal Component Analysis (ALPCA), which is able to handle sudden as well as gradual changes in background. Next, the extracted human silhouettes are used to train a appearance-based algorithm that can recognize the human poses. The excellent results reported here justify the elimination of active sensors and game controllers for many traditional games.

The “killer apps” of cellular and swarm computing are image processing and optimization, respectively; however, applying these platforms to general-purpose computing remains impractical. Designing systems within the restrictive framework of cellular automata is extremely difficult, though often very efficient and scalable. On the other hand, swarm networks are very powerful but difficult to implement in hardware. Here we introduce a hybrid model, the Swarm Computer, which is both practical to program and efficient to implement. Applications in astrophysics and image processing are considered.

In this paper, a homography-based approach for determining the ground plane using image pairs is presented. Our approach is unique in that it uses a Modified Expectation Maximization algorithm to cluster pixels on images as belonging to one of two possible classes: ground and non-ground pixels. This classification is very useful in mobile robot navigation because, by segmenting out the ground plane, we are left with all possible objects in the scene, which can then be used to implement many mobile robot navigation algorithms such as obstacle avoidance, path planning, target following, landmark detection, etc. Specifically, we demonstrate the usefulness and robustness of our approach by applying it to a target following algorithm. As the results section shows, the proposed algorithm for ground plane detection achieves an almost perfect detection rate (over 99%) despite the relatively higher number of errors in pixel correspondence from the feature matching algorithm used: SIFT.

We present an efficient path planner for smart wheelchairs based on harmonic potential fields. While the use of harmonic fields can always guarantee finding an existing path, they are extremely computational intensive and a sufficiently detailed map of the environment may lead to an unfeasible solution for the path. Also, since our target application is for the navigation of a smart wheelchair, for people with severe disabilities, the path provided by the harmonic field is frequently too sharp and needs to be smoothened. In order to address the first problem, we propose a parallel algorithm implemented using Graphics Processor Units (GPUs) on the Compute Unified Device Architecture (CUDA) platform. And for the second problem, we developed a rubber band model that provides extra forces to be added to the attracting forces of the harmonic fields. This model assumes that the path is an elastic line, a rubber band, connecting the source and destination points. This rubber band simulates the internal tension forces trying to tighten the line. As the result section demonstrates, both the original path from the harmonic field alone and the path smoothened by the rubber band model have approximate the same length, but the first path contains many bumps, sharp angles, and zig-zags, while the second one provides a much more comfortable ride for the passenger of the wheelchair. Either one is executed in real-time, allowing the proposed method to be used for real navigation of smart wheelchairs.

We propose a robust and accurate method for multi-target geo-localization from airborne video. The difference between our approach and other approaches in the literature is fourfold: 1) it does not require gimbal control of the camera or any particular path planning control for the UAV; 2) it can instantaneously geo-locate multiple targets even if they were not previously observed by the camera; 3) it does not require a geo-referenced terrain database nor an altimeter for estimating the UAV's and the target's altitudes; and 4) it requires only one camera, but it employs a multi-stereo technique using the image sequence for increased accuracy in target geo-location. The only requirements for our approach are: that the intrinsic parameters of the camera be known; that the on board camera be equipped with global positioning system (GPS) and iner-tial measurement unit (IMU); and that enough feature points can be extracted from the surroundings of the target. Since the first two constraints are easily satisfied, the only real requirement is regarding the feature points. However, as we explain later, this last constraint can also be alleviated if the ground is approximately planar. The result is a method that can reach a few meters of accuracy for an UAV flying at a few hundred meters above the ground. Such performance is demonstrated by computer simulation, in-scale data using a model city, and real airborne video with ground truth.

The inherent massive parallelism of cellular neural networks makes them an ideal computational platform for kernel-based algorithms and image processing. General-purpose GPUs provide similar massive parallelism, but it can be difficult to design algorithms to make optimal use of the hardware. The presented research includes a GPU abstraction based on cellular neural networks. The abstraction offers a simplified view of massively parallel computation which remains reasonably efficient. An image processing library with visualization software has been developed to showcase the flexibility and power of cellular computation on GPUs. Benchmarks of the library indicate that commodity GPUs can be used to significantly accelerate CNN research and offer a viable alternative to CPU-based image processing algorithms.

We propose a robust and accurate method for multi-target geo-localization from airborne video. The difference between our approach and other approaches in the literature is fourfold: 1) it does not require gimbal control of the camera or any particular path planning control for the UAV; 2) it can instantaneously geolocate multiple targets even if they were not previously observed by the camera; 3) it does not require a geo-referenced terrain database nor an altimeter for estimating the UAV's and the target's altitudes; and 4) it requires only one camera, but it employs a multi-stereo technique using the image sequence for increased accuracy in target geo-location. The only requirements for our approach are: that the intrinsic parameters of the camera be known; that the on board camera be equipped with global positioning system (GPS) and inertial measurement unit (IMU); and that enough feature points can be extracted from the surroundings of the target. Since the first two constraints are easily satisfied, the only real requirement is regarding the feature points. However, as we explain later, this last constraint can also be alleviated if the ground is approximately planar. The result is a method that can reach a few meters of accuracy for an UAV flying at a few hundred meters above the ground. Such performance is demonstrated by computer simulation, in-scale data using a model city, and real airborne video with ground truth.

This paper presents a method for multi-view 3D modeling of human bodies using virtual stereopsis. The algorithm expands and improves the method used in, but unlike that method, our approach does not require multiple calibrated cameras and/or carefully-positioned turn tables. Instead, an algorithm using SIFT feature extraction is employed and an accurate motion estimation is performed to calculate the position of virtual cameras around the object. That is, by employing a single pair of cameras mounted on a same tripod, our algorithm computes the relative pose between camera and object and creates virtual cameras from the consecutive images in the video sequence. Besides not requiring any special setup, another advantage of our method is in the simplicity to obtain denser models if necessary: by only increasing the number of sampled images during the object-camera motion. As the quantitative results presented here demonstrate, our method compares to the PMVS method, while it makes it much simpler and cost-effective to implement.

This paper presents a new 3D representation for non-rigid objects using motion vectors between two consecutive frames. Our method relies on an Octree to recursively partition the object into smaller parts for which a small number of motion parameters can accurately represent that portion of the object. The partitioning continues as long as the respective motion parameters are insufficiently accurate to describe the object. Unlike other Octree methods, our method employs an affine transformation for the motion description part, which greatly reduces the storage. Finally, an adaptive thresholding, a singular value decomposition for dealing with singularities, and a quantization and arithmetic coding further enhance our proposed method by increasing the compression while maintaining very good signal-noise ratio. Compared with other methods like tri-linear interpolation or Principle Component Analysis (PCA) based algorithm, the Affine-Octree method is easy to compute and highly compact. As the results demonstrate, our method has a better performance in terms of compression ratio and PSNR, while it remains simple.

This paper presents two different algorithms for object tracking and pose estimation. Both methods are based on an appearance model technique called Active Appearance Model (AAM). The key idea of the first method is to utilize two instances of the AAM to track landmark points in a stereo pair of images and perform 3D reconstruction of the landmarks followed by 3 D pose estimation. The second method, the AAM matching algorithm is an extension of the original AAM that incorporates the full 6 DOF pose parameters as part of the minimization parameters. This extension allows for the estimation of the 3D pose of any object, without any restriction on its geometry. We compare both algorithms with a previously developed algorithm using a geometric-based approach [14]. The results show that the accuracy in pose estimation of our new appearance-based methods is better than using the geometric-based approach. Moreover, since appearance-based methods do not require customized feature extractions, the new methods present a more flexible alternative, especially in situations where extracting features is not simple due to cluttered background, complex and irregular features, etc.

We present a distributed vision-based architecture for smart robotics that is composed of multiple control loops, each with a specialized level of competence. Our architecture is subsumptive and hierarchical, in the sense that each control loop can add to the competence level of the loops below, and in the sense that the loops can present a coarse-to-fine gradation with respect to vision sensing. At the coarsest level, the processing of sensory information enables a robot to become aware of the approximate location of an object in its field of view. On the other hand, at the finest end, the processing of stereo information enables a robot to determine more precisely the position and orientation of an object in the coordinate frame of the robot. The processing in each module of the control loops is completely independent and it can be performed at its own rate. A control Arbitrator ranks the results of each loop according to certain confidence indices, which are derived solely from the sensory information. This architecture has clear advantages regarding overall performance of the system, which is not affected by the "slowest link," and regarding fault tolerance, since faults in one module does not affect the other modules. At this time we are able to demonstrate the utility of the architecture for stereoscopic visual servoing. The architecture has also been applied to mobile robot navigation and can easily be extended to tasks such as "assembly-on-the-fly."

This paper presents a fast tracking algorithm capable of estimating the complete pose (6DOF) of an industrial object by using its circular-shape features. Since the algorithm is part of a real-time visual servoing system designed for assembly of automotive parts on-the-fly, the main constraints in the design of the algorithm were: speed and accuracy. That is: close to frame-rate performance, and error in pose estimation smaller than a few millimeters. The algorithm proposed uses only three model features, and yet it is very accurate and robust. For that reason both constraints were satisfied: the algorithm runs at 60 fps (30 fps for each stereo image) on a PIII-800 MHz computer, and the pose of the object is calculated within an uncertainty of 2.4 mm in translation and 1.5 degree in rotation.

This paper describes an approach for calibrating active stereo heads for which no prior knowledge of the world coordinate system is required. During an automated procedure derived from this approach, the mobile robot can be loosely positioned in front of a calibration pattern. Through our procedure, both the camera calibration and the neck-eye calibration are obtained. With this set of calibration matrices, the XYZ coordinates of a landmark can be determined with respect to the robot coordinate frame. The so calculated XYZ coordinates are accurate regardless of the orientation of the stereo head, even for orientations that were not used for calibration. This paper validates our approach by presenting results as accurate as a few millimeters in the calculation of the XYZ coordinates of objects at 1.5m from the stereo head.

Surveys the developments of the last 20 years in the area of vision for mobile robot navigation. Two major components of the paper deal with indoor navigation and outdoor navigation. For each component, we have further subdivided our treatment of the subject on the basis of structured and unstructured environments. For indoor robots in structured environments, we have dealt separately with the cases of geometrical and topological models of space. For unstructured environments, we have discussed the cases of navigation using optical flows, using methods from the appearance-based paradigm, and by recognition of specific objects in the environment.

In visual servoing applications using a position-based approach and an end-effector-mounted camera, the position and orientation of the camera with respect to the end-effector must be known. This information is frequently represented in the form of a homogeneous transformation matrix (HTM). For special "noise-free" cases, a closed-form solution for this calibration problem can be determined. However, in the real world, such a solution is not adequate and a least-squares approach or an adaptive algorithm must be used. In this paper, we describe a new algorithm that can simultaneously calculate the base-world and the hand-eye (camera to end-effector) HTMs. This method is robust to noise, and converges to a valid solution very quickly.

Robotic systems and their software design are based on the same principles found in other computer applications. However, while many computer systems to date are deeply rooted in concepts such as multi-processing, multi-threads, modularity, etc., robotic systems are still limited by mono-processing, centralized designs. The paper addresses these limitations by proposing a new philosophy for robotic system software design - software infrastructure $that allows for the design of an efficient, modular, fault-tolerant, and distributed software architecture. Two examples of this infrastructure applied to mobile robot navigation are also discussed.

The above paper by Pan et al. (ibid. vol.6 (1998)) presents a rather complicated mechanism allowing fuzzy evidence to be aggregated when fuzzy inferences are made about the same fuzzy variable by different rules. The purpose of this comment is to present a simpler way to address this FuzzyCLIPS problem. In reply, the authors point out that the commenter's main conclusions about their work are ill founded. They believe that his erroneous conclusions were caused by his misunderstanding the difference between rule firing and evidence aggregation in their work.

There exist in the literature today many contributions dealing with the incorporation of fuzzy logic in expert systems. However, unfortunately, much of what has been proposed can only be applied to small-scale expert systems; that is, when the number of rules is in the dozens as opposed to in the hundreds. The more traditional (nonfuzzy) expert systems are able to cope with large numbers of rules by using Rete networks for maintaining matches of all the rules and all the facts. (A Rete network obviates the need to match the rules with the facts on every cycle of the inference engine.) In this paper, we present a more general Rete network that is particularly suitable for reasoning with fuzzy logic. The generalized Rete network consists of a cascade of three networks: the pattern network, the join network, and the evidence aggregation network. The first two layers are modified versions of similar layers for the traditional Rete networks and the last, the aggregation layer, is a new concept that allows fuzzy evidence to be aggregated when fuzzy inferences are made about the same fuzzy variable by different rules.