An application of Lyapunov stability analysis to improve the performance of NARMAX models

Previously we presented a novel approach to program a robot controller based on system identification and robot training techniques. The proposed method works in two stages: first, the programmer demonstrates the desired behaviour to the robot by driving it manually in the target environment. During this run, the sensory perception and the desired velocity commands of the robot are logged. Having thus obtained training data we model the relationship between sensory readings and the motor commands of the robot using ARMAX/NARMAX models and system identification techniques. These produce linear or non-linear polynomials which can be formally analysed, as well as used in place of “traditional robot” control code.In this paper we focus our attention on how the mathematical analysis of NARMAX models can be used to understand the robot’s control actions, to formulate hypotheses and to improve the robot’s behaviour. One main objective behind this approach is to avoid trial-and-error refinement of robot code. Instead, we seek to obtain a reliable design process, where program design decisions are based on the mathematical analysis of the model describing how the robot interacts with its environment to achieve the desired behaviour. We demonstrate this procedure through the analysis of a particular task in mobile robotics: door traversal.     

Ensemble cluster pruning via convex-concave programming

Ensemble learning is the process of aggregating the decisions of different learners/models. Fundamentally, the performance of the ensemble relies on the degree of accuracy in individual learner predictions and the degree of diversity among the learners. The trade-off between accuracy and diversity within the ensemble needs to be optimized to provide the best grouping of learners as it relates to their performance. In this optimization theory article, we propose a novel ensemble selection algorithm which, focusing specifically on clustering problems, selects the optimal subset of the ensemble that has both accurate and diverse models. Those ensemble selection algorithms work for a given number of the best learners within the subset prior to their selection. The cardinality of a subset of the ensemble changes the prediction accuracy. The proposed algorithm in this study determines both the number of best learners and also the best ones. We compared our prediction results to recent ensemble clustering selection algorithms by the number of cardinalities and best predictions, finding better and approximated results to the optimum solutions.     

Production of titanium-containing metal-ceramic composites based on boron carbide in the nanocrystalline state

ABSTRACT

The results of the study of the production technology, phase composition, structure and physico-mechanical properties of metal-ceramic materials based on boron carbide and their components are presented. Boron carbide was obtained by direct synthesis from chemical elements using amorphous boron and carbon black. By mechanical dispersion, solid reagents were converted into an ultrafine state. Using a chemical method, nanoscale (70–80?nm) boron carbide was synthesised from suspension solutions of amorphous boron and liquid hydrocarbons. Boron carbide-based metal-ceramic composite powder B₄C–(Co–Ni–Ti) was obtained by mechanical dispersion of the constituent components. Based on results of studying of the temperature-dependence of wetting angle of boron carbide with Co–Ni–Ti metallic alloy, the compacting modes of metal-ceramic composite powders by plasma-spark sintering and hot pressing have been developed. The influence of the component content of the binder metal (alloy) on some physico-mechanical properties (linear expansion coefficient, hardness, and bending strength) of hardmetal-ceramic materials based on boron carbide was studied. It was found that the optimum content of the metal component in the composite is ～ 25?wt-%. In the temperature range 300–600°C, the materials obtained are characterised by stable dimensional factors, since in this temperature range the thermal conductivity coefficient does not depend much on temperature. At room temperature, their bending strength is about 1?GPa. A new method of chemical synthesis of nanocrystalline ceramic compositions of boron carbide and titanium diboride using suspension solutions for the preparation of powders and their spark plasma sintering was also developed to obtain a compacted material of composition B₄C+30?wt-%TiB₂, which has a high hardness of 95 HRA (with maximum microhardness 45.6?GPa) and sufficient strength (with a bending strength of 834?MPa).     

Towards modelling complex robot training tasks through system identification

Previous research has shown that sensor–motor tasks in mobile robotics applications can be modelled automatically, using NARMAX system identification, where the sensory perception of the robot is mapped to the desired motor commands using non-linear polynomial functions, resulting in a tight coupling between sensing and acting — the robot responds directly to the sensor stimuli without having internal states or memory.However, competences such as for instance sequences of actions, where actions depend on each other, require memory and thus a representation of state. In these cases a simple direct link between sensory perception and the motor commands may not be enough to accomplish the desired tasks. The contribution of this paper to knowledge is to show how fundamental, simple NARMAX models of behaviour can be used in a bootstrapping process to generate complex behaviours that were so far beyond reach.We argue that as the complexity of the task increases, it is important to estimate the current state of the robot and integrate this information into the system identification process. To achieve this we propose a novel method which relates distinctive locations in the environment to the state of the robot, using an unsupervised clustering algorithm. Once we estimate the current state of the robot accurately, we combine the state information with the perception of the robot through a bootstrapping method to generate more complex robot tasks: We obtain a polynomial model which models the complex task as a function of predefined low level sensor–motor controllers and raw sensory data.The proposed method has been used to teach Scitos G5 mobile robots a number of complex tasks, such as advanced obstacle avoidance, or complex route learning.     

A fish perspective: detecting flow features while moving using an artificial lateral line in steady and unsteady flow

For underwater vehicles to successfully detect and navigate turbulent flows, sensing the fluid interactions that occur is required. Fish possess a unique sensory organ called the lateral line. Sensory units called neuromasts are distributed over their body, and provide fish with flow-related information. In this study, a three-dimensional fish-shaped head, instrumented with pressure sensors, was used to investigate the pressure signals for relevant hydrodynamic stimuli to an artificial lateral line system. Unsteady wakes were sensed with the objective to detect the edges of the hydrodynamic trail and then explore and characterize the periodicity of the vorticity. The investigated wakes (Kármán vortex streets) were formed behind a range of cylinder diameter sizes (2.5, 4.5 and 10 cm) and flow velocities (9.9, 19.6 and 26.1 cm s⁻¹). Results highlight that moving in the flow is advantageous to characterize the flow environment when compared with static analysis. The pressure difference from foremost to side sensors in the frontal plane provides us a useful measure of transition from steady to unsteady flow. The vortex shedding frequency (VSF) and its magnitude can be used to differentiate the source size and flow speed. Moreover, the distribution of the sensing array vertically as well as the laterally allows the Kármán vortex paired vortices to be detected in the pressure signal as twice the VSF.     

Hybrid sorbent-ultrafiltration systems for fluoride removal from water

Fluoride contaminated water sources are found in many parts of the world and the consumption of such water is causing dental and skeletal fluorosis in humans, especially in developing countries. Hybrid sorbent-ultrafiltration (UF) systems are proposed for the removal of fluoride from water for the first time in this study. Laterite and bone char were selected as they are low cost, accessible sorbents in developing countries. The performances of the laterite-UF and bone char-UF systems were compared in terms of fluoride removal and membrane permeability under varying fluoride concentration, solution pH, and sorbent load. For equilibrium fluoride concentration of 1.5 mg/L, the World Health Organization guideline for safe drinking water, the sorption capacity of bone char (1.1 mg/g) was larger than that of laterite (0.40 mg/g) and this was attributed to the larger surface area of bone char. For the laterite-UF system, increase in fluoride concentration resulted in a decline in UF permeability whereas for the bone char-UF system there was no influence of fluoride concentration on membrane permeability. The optimal solution pH at which the systems are operated at maximum sorption capacity while avoiding membrane fouling was determined as pH 5-6 for the laterite-UF and pH 7 for the bone char-UF system. For both systems, the permeability declined in a similar manner as the sorbent load increased. Although both systems require further optimization, they showed to be viable defluoridation technologies.