Uri Ben-Hanan

Modeling of caterpillar crawl using novel tensegrity structures

Caterpillars are soft-bodied animals. They have a relatively simple nervous system, and yet are capable of exhibiting complex movement. This paper presents a 2D caterpillar simulation which mimics caterpillar locomotion using Assur tensegrity structures. Tensegrity structures are structures composed of a set of elements always under compression and a set of elements always under tension. Assur tensegrities are a novel sub-group of tensegrity structures. In the model, each caterpillar segment is represented by a 2D Assur tensegrity structure called a triad. The mechanical structure and the control scheme of the model are inspired by the biological caterpillar. The unique engineering properties of Assur tensegrity structures, together with the suggested control scheme, provide the model with a controllable degree of softness-each segment can be either soft or rigid. The model exhibits several characteristics which are analogous to those of the biological caterpillar. One such characteristic is that the internal pressure of the caterpillar is not a function of its size. During growth, body mass is increased 10 000-fold, while internal pressure remains constant. In the same way, the model is able to maintain near constant internal forces regardless of size. The research also suggests that caterpillars do not invest considerably more energy while crawling than while resting.

Control of irrigation amounts using velocity and position of wetting front

A new approach for the estimation and control of the quantity of water applied in an irrigation is presented in which irrigation is stopped when the wetting front reaches a critical depth, ZL. An expression for calculating the critical depth ZL was developed. A major parameter in this expression is the velocity of advance of the wetting front, V, which was shown to be directly related to the application rate, IR, and inversely related to the initial soil water content, θi. A depth probe (patent pending) was designed, constructed and tested for the purpose of monitoring the position of the wetting front during infiltration and redistribution and for computing the value of V. Equations developed for relating the velocity of advance of the wetting front to θi as well as for estimating the value of the critical depth ZL were successfully tested under conditions of uniform distribution of the initial soil water content. An iterative learning process which utilizes the real time output from the depth probe during each irrigation and is therefore capable of handling realistic field conditions where nonuniformity is the rule is presented. The acquired information is used to estimate a critical depth of the wetting front, ZL, for a planned final wetted depth, ZF, during each irrigation. This process is incorporated in the depth probe and is used to stop irrigation and thus control the quantity of water applied.

Classification of fruits by a Boltzmann perceptron neural network

Abstract Classification of fruits by machine vision is problematic in two respects: (a) Most of the sorting criteria are “fuzzy”, because the class membership can not be quantified precisely. The reference classification is subjectively determined by a trained panel of inspectors, that often disagree as to the class of the fruit. (b) The statistics of the classification criteria vary with harvest time and from orchard to orchard, so the classifier must be easy to re-train. Using digital color imaging hardware and a BPN based classifier we developed a system for sorting fruits that can address these problems. It naturally accepts “fuzzy” or “soft” labeling at train-time and can be tuned to provide various levels “soft” or “fuzzy” decisions at run-time. The power of the BPN was demonstrated by a synthetic dataset, indicating that the BPN can create intricate non-linear discriminant functions, even when the classes are noncontinguous and the training set is relatively small. Simulated sorting experiments of apples, into “red” and “green” categories showed that the system can emulate the decisions of a panel of human sorters quite well.

A MODEL OF CATERPILLAR LOCOMOTION BASED ON ASSUR TENSEGRITY STRUCTURES

This paper presents an ongoing project aiming at building a robot composed of Assur tensegrity structures, which mimics caterpillar locomotion. Caterpillars are soft-bodied animals capable of making complex movements with astonishing fault-tolerance. In our model, each caterpillar segment is represented by a 2D tensegrity triad consisting of two bars connected by two cables and a strut. The cables represent the major longitudinal muscles of the caterpillar, while the strut represents hydrostatic pressure. The control scheme in this model is divided into localized low-level controllers and a high-level control unit. The unique engineering properties of Assur tensegrity structures, which were mathematically proved last year, together with the suggested control algorithm provide the model with robotic softness. Moreover, the degree of softness can be continuously changed during simulation, making this model suitable for simulation of soft-bodied caterpillars as well as other types of soft animals.Copyright

ADJUSTABLE TENSEGRITY ROBOT BASED ON ASSUR GRAPH PRINCIPLE

The paper introduces a tensegrity robot consisting of cables and actuators. Although this robot has zero degrees of freedom, it is both mobile, and capable of sustaining massive external loads. This outcome is achieved by constantly maintaining the configuration of the robot at a singular position. The underlying theoretical foundation of this work is originated from the concept of Assur Trusses (also known as Assur Groups), which are long known in the field of kinematics. During the last three years, the latter concept has been reformulated by mathematicians from rigidity theory community, and new theorems and algorithms have been developed. Since the topology of the robot is an Assur Truss, the work reported in the paper relies on Assur Trusses theorems that have been developed this year resulting in an efficient algorithm to constantly keep the robot at the singular position. In order to get an efficient characterization of the desired configurations, known techniques from projective geometry were employed. The main idea of the control system of the device, that was also mathematically proved, is that changing the length of only one element, causes the robot to be at the singular position. Therefore, the system measures the force in only one cable, and its length is modified accordingly by the control system. The topology of the device is an Assur Truss — a 3D triad, but the principles introduced in the paper are applicable to any robot whose topology is an Assur Truss, such as: tetrad, pentad, double triad and so forth. The paper includes several photos of the device and the output data of the control system indicating its promising application.Copyright

ADAPTIVE CLASSIFICATION BY NEURAL NET BASED PROTOTYPE POPULATIONS

We have developed an algorithm for unsupervised adaptive classification based on a finite number of “prototype populations” with distinctly different feature distributions, each representing a typically different source population of the inspected products. Intermittently updated feature distributions, of samples collected from the currently classified products, are compared to the distributions of pre-stored prototype populations, and accordingly the system switches to the most appropriate classifier. The goal of our approach is similar to the objectives of the previously proposed “Decision Directed” adaptive classification algorithms but our solution is particularly suitable for automatic inspection and classification on a production line, when the inspected items may come from a finite number of distinctly different sources. The recognition of prototype populations as well as the classification task proper may be implemented by conventional classifiers, however neural networks (NN) are advantageous in two respects: There is no need to develop separate mathematical models for each classifier because the NN does it automatically during the training stage. The parallel structure of NN has the potential for very fast classification in real time, if implemented by dedicated parallel hardware. This is particularly important for high speed automatic sorting on a production line. The practical feasibility of the approach was demonstrated by two applied examples, wherein two prototype populations of apples are recognized and sorted by size and color derived by machine vision. Three “Boltzmann-Perceptron Networks” (BPN) were used, one to recognize the prototype populations, while switching between the other two, for optimally classifying apples into two size and color categories. It is shown that misclassifications by adaptive classification are reduced, in comparison to non-adaptive classification.

Classification of apples with a neural network based classifier

Abstract 352 apples were classified as green or red by three human experts. Each apple got a probabilistic (fuzzy) membership in the red and green sets according to the number of votes. The majority decided the actual classification. This fuzzy data, and features such as color hue, saturation, and intensity, and the ratio between red and green areas, for the first 30 (or 60) apples were used to train a Boltzmann Perceptron Network (BPN). The remaining fruits were then classified by the BPN. It was found that hue is the dominant feature. The misclassification was less than 15%, similar to a Bayesian classifier trained on the same data. The advantages of the BPN seem to be that it allows fuzzy input data, and many features.

Comparative study of three different types of dental diamond burs

Abstract Wear mechanisms of three different types of dental burs were studied by means of cutting experiments performed on machinable glass ceramic using a laboratory system designed for this purpose. The dental handpiece used for this research was subjected to a constant feed rate in order to better simulate the actual working conditions of a dental bur. The new and the worn-out burs were studied by optical and scanning electron microscopy. Diamond particle wear-out was found to be the dominant wear mechanism in all cases; a quantitative analysis was performed on the optical micrographs of the new and worn burs. In situ force measurements showed that the forces exerted by the bur increase with the blunting process in order to keep the required feed rates; each bur type seems to have a different characteristic curve of force versus the number of cuts.

Adaptive sorting by prototype populations

Abstract An algorithm for unsupervised adaptive sorting is presented, based on a finite number of ‘prototype populations’, with distinctly different feature distributions, each representing a typically different source population of the inspected products. Updated feature distributions, of samples collected from the currently sorted products, are compared to the distributions of the stored prototype populations, and accordingly the system switches to the most appropriate classifier. Although the goal is similar to the objectives of previously proposed ‘Decision Directed’ adaptive classification algorithms, the present algorithm is particularly suitable for automatic inspection and classification on a production line, when the inspected items may come from different sources. The practical feasibility of the approach is demonstrated by two synthetic examples, using Bayes classifiers. This is followed by an applied example, wherein two prototype populations of apples are sorted by size, derived by machine vision. It is shown that misclassification by adaptive classification is reduced, in comparison to non-adaptive classification.

Classification by varying features with an erring sensor

A method is proposed for unsupervised classification by a feature that may vary with time, measured by an erring sensor. A classification threshold for the erring sensor is found such that the misclassification is minimized. It is shown that the method is an application of Bayes rule without knowledge of the a priori probabilities, while estimating the class conditional probabilities by an erring sensor model. Sorting of fruits is presented as an illustrative example.