Institute of Fundamental Technological Research

Arm-Z to koncepcja hiperredundantnego manipulatora robotycznego opartego na sekwencji szeregowo połączonych identycznych modułów. Każdy moduł ma tylko jeden stopień swobody (1-DOF) – skręt względem poprzedniego. Moduły systemu Arm-Z mogą być masowo produkowane i łatwo wymieniane w przypadku awarii. Sterowanie Arm-Z jest stosunkowo trudne, dlatego zwykle wymaga stosowania metod inteligencji obliczeniowej. W artykule przedstawiono kilka koncepcji kinetycznych obiektów małej architektury opartych na Arm-Z: spiralną kolumnę o regulowanej wysokości, system nadążania słonecznego, kinetyczną rzeźbę bioniczną i kinetyczny zraszacz/fontannę. Prezentowane koncepcje są zasadniczo nisko-technologiczne (“low-tech”). W każdym przypadku moduł bazowy jest przymocowany do podstawy (podłoża). Dla prostoty napęd jest przykładany bezpośrednio do pierwszego modułu, a następnie przenoszony za pomocą wewnętrznych przekładni na kolejne moduły. Każdy moduł jest wyposażony w zestaw cylindrycznych i stożkowych kół zębatych z zębami prostymi o profilu spiralnym (do połączeń miedzy modułami).

Arm-Z, hiperredundancja, manipulator, mała architektura, systemy modularne

In order to perform mapping, inspecting, searching, painting, cleaning, and other similar tasks, mobile robots have to act according to a coverage plan. Finding a trajectory that a robot should follow requires an appropriate coverage path planning (CPP) algorithm and is a non-trivial problem, especially if a cooperating group of robots is considered. We propose that the multi-robot CPP can be solved by: decomposing the input occupancy grid map into cells, generating a corresponding Reeb graph, clustering the graph into Nr clusters, and solving the associated equality generalized traveling salesman problem in order to obtain optimal back-and-forth sweeping patterns on the clusters. This last step has been proven to be one of the most efficient ways to find trajectories for a single robot [5]. The discussed approach is motivated by a specific application: industrial cleaning of large warehouses by Nr autonomic mobile cleaners (the cleaning radius of a robot is much smaller than the area to be cleaned). The total time required for cleaning is to be minimized. By means of statistical analysis, using an extensive, realistic set of synthetic maps, it is shown that the proposed algorithm meets the criteria for applying it in the production process.

coverage path planning,motion planning,multi-robot

Arm-Z - a hyper-redundant manipulator based on linearly joined sequence of congruent units is presented. Each unit has 1-DOF (one degree of freedom) only, namely a twist relative to the previous unit in the sequence. Arm-Z has a potential of being economical and robust. The control of Arm-Z, however, is not intuitive and difficult. This paper presents the preliminary results of designing a prototype of a modular Sun-tracking device comprised of four congruent units with possible application as a solar energy harvester or a Sun-shade.

Extremely modular system, Arm-Z, Hyper-redundant manipulator, Rapid prototyping, Low-tech

Arm-Z is a hyper-redundant manipulator based on a sequence of linearly joined identical modules. Each module has only one degree of freedom—a twist relative to the previous module. Arm-Z can be potentially economical, as the modules can be mass-produced. Arm-Z is also robust, as the malfunctioning module can be replaced. Moreover, if some modules malfunction, the device can still execute tasks with certain accuracy. However, the disadvantage of Arm-Z is a non-intuitive and difficult control. This paper presents a concept of a modular tracking device comprised of four identical modules. As an example, the Sun-tracking setup is used with possible application for solar energy harvesting.

Extremely modular system, Arm-Z, Hyper-redundant manipulator, Sun-tracking

Improving the architectural layout for diverse objectives using rigorous mathematical optimization methods gradually receives more attention by the researchers. Such optimization however, is usually reduced to a much simpler and relatively well-defined problem such as: facility layout optimization, quadratic assignment problem, rectangle partitioning. Nonetheless, architects are usually skeptical about such approaches since they produce solutions which lack certain architectural qualities. This paper proposes a framework where architectural functional layout (FL) is optimized for the following objectives: functionality (defined by users), insolation (calculated according to geographical conditions), outside view attractiveness (assessed on-site) and external noise (measured on-site). Incorporating the latter two and simultaneous optimization of FLs for objectives related specifically to the site: position and orientation are the novel contributions of this paper. Firstly, a set of candidate FLs is generated, next they are evaluated for optimal location and orientation on a given site. Optimality is conceived here as maximization of real-valued objective function combining: user's satisfaction level of the outside views, shielding from external noise, and insolation
preference. The importance of these factors for each type of room is assessed by the user (as weights). A case study on an existing site is presented. The view quality was arbitrarily assessed and the noise map was assessed by A-weighted equivalent sound level measurements. A general gradient-based method for finding optimal and near-optimal solutions was applied. The output of this optimization is a set of room configurations with their locations and orientations on the site returned to the user for final selection.

architectural optimization, subjective evaluation, functional layout, acoustic comfort, coarse grid

We apply a non-classical four-valued logic in the process of reasoning regarding strategies for cops in a modified game of "Cops and Robber" played on a graph. We extend the game by introducing uncertainty in a form of random failures of detecting devices. This is realized by allowing that a robber can be detected in a node only with the given probability PA. Additionally, with the probability PF, cops can be given a false-positive, i.e., they are informed that the robber is located at some node, whereas it is located somewhere else. Consequently, non-zero PF introduces a measurement noise into the system. All the cops have access to information provided by the detectors and can communicate with each other, so they can coordinate the search. By adjusting the number of detectors, PA, and PF we can achieve a smooth transition between the two well-known variants of the game: "with fully visible robber" and "with invisible robber". We compare a simple probabilistic strategy for cops with the non-parametric strategy based on reasoning with a four-valued paraconsistent logic. It is shown that this novel approach leads to a good performance, as measured by the required mean catch-time. We conclude that this type of reasoning can be applied in real-world applications where there is no knowledge about the underlying source of errors which is particularly useful in robotics.

reasoning under uncertainty, non-classical logics, four-valued logic, paraconsistent reasoning, cops and robber game, multi-robotic systems

Truss-Z (TZ) is an Extremely Modular System for creating skeletal free-form ramps and ramp networks. The TZ structures are comprised of four variations of two types of basic unit subjected to rotation. The two types of units are: R and L being a mirror reflection of each other. This paper presents a novel method based on image processing, evolutionary algorithm and intensive parallelization of multi-objective optimization of TZ layouts.
The algorithm returns a sequence of modules. The result guarantees a TZ connection between two given points (regions) and minimizes the fitness function representing certain costs associated with setting up the TZ structure.
The fitness function depends on the cost of TZ structure as well as the variety of costs related to the environment where the it is to be placed. E.g.: the earthworks, vegetation removal, obstacles avoidance, etc. There are no restrictions on the fitness function definition. It can depend on any variable which can be represented by a two-dimensional map of any property of the environment.
The formulation of the presented method is suited for application of well-established image processing methods which efficiently evaluate candidate solutions on a GPU. As a result, the employed genetic algorithm efficiently probes the search space. The practical applicability of this approach is demonstrated with three case-studies:
1) simultaneous paving of a path with congruent units in a hilly environment with trees & bushes and finding the best location for a pier over an existing river;
2) constructing of a TZ connector spanning over a mountain valley with lakes (where supports can not be placed);
3) retrofitting of an existing railway station with a large wheelchair TZ ramp of over 10 m elevation while preserving trees and minimizing the earthworks.

Truss-Z, Extremely Modular System, Retrofitting, accessibility, multi-objective, discrete, combinatorial, optimization, genetic algorithm, parallel computing, GPU, GPGPU

A novel concept of hyper-redundant, snake-like manipulator is presented. It is based on the reconfigurable modular construction system–Arm-Z (AZ). AZ is comprised of linearly joined congruent modules with possibility of relative twist. AZ is an Extremely Modular System, i.e. it is composed of a single basic unit and allows for creating free-form shapes. Required level of usefulness and efficiency are among the most challenging design aspects of such reconfigurable systems. Here AZ is considered in the context of kinematics of robotic arms. In general, due to its highly non-linear nature, it is very difficult to find transitions between given states (configurations), especially under realistic environmental and structural constraints. As a way to control the manipulator, an implementation of Particle Swarm Optimization (PSO) for finding transitions between AZ states in realistic scenarios is proposed. Four practical examples are presented which are variations of two distinct problems: bending of a hexagonal AZ in a narrow slot (strong environmental constraints), and reaching a given point in 3D space by the tip of dodecagonal AZ (acting as a robotic arm). The problem of AZ transformation has been defined as a multi-objective optimization. The methodology is general with no restrictions to the objective function. Since the problem is strongly non-linear, in order to cover large space of potential solutions, the algorithm runs for a relatively large number of random initial swarms. This task was distributed on a computer cluster. Although the nature of AZ reconfiguration is discrete, the optimization algorithm is continuous.

Extremely modular system, Arm-Z, Pipe-Z, Discrete optimization, Dihedral rotation, Reconfiguration, Particle Swarm Optimization, Redundant robot, Hyper-redundant manipulator

Niniejszy artykuł opisuje prace wykonane w celu przeprowadzenia eksperymentu weryfikującego działanie Planera zadania Search and Rescue na zespole trzech robotów kołowych w budynku użyteczności publicznej. Celem zadania jest przeszukanie obiektu przez grupę robotów, odnalezienie w nim poszkodowanych oraz wezwanie pomocy w odpowiednie miejsce. Punktem centralnym systemu jest Planer wykorzystujący język logiki czterowartosciowej 4QL, który został uprzednio przetestowany na symulacjach komputerowych. Jego działanie zostanie opisane pobieżnie, głównym tematem jest metodyka wdrożenia. Aby przeprowadzić eksperyment, należało zintegrować Planer z systemem wielorobotowym. Wedle autorów jest to zadanie nietrywialne, dlatego przy jego realizacji wykorzystano paradygmaty Test-driven development (TDD), zaadaptowane odpowiednio do realiów robotyki mobilnej. Finalny eksperyment został nagrany, a wideo udostępniono w Internecie.

logika wielowartościowa, search and rescue, systemy wielorobotowe

Jedną z zalet systemów składających się z wielu robotów mobilnych jest możliwość współdziałania jednostek w celu zwiększenia dokładności ich samolokalizacji. W artykule omówiono sytuację, w której zakłada się że roboty praktycznie nie mają dostępu do czujników pozycji bezwzględnej i swoją pozycję ustalają jedynie na podstawie czujników proprioceptywnych, wzajemnych obserwacji oraz -- opcjonalnie -- obserwacji pojedynczego znacznika, co umożliwia osadzenie pozycji w układzie globalnym. Na przykładzie problemu kooperatywnego przeszukania obszaru, poprzez przeprowadzenie symulacji, pokazano że odpowiednie zaplanowanie trajektorii pozwala w istotnym stopniu zwiększyć dokładność pozycjonowania.

planowanie ścieżki, systemy wielorobotowe, wzajemne obserwacje

In the paper we demonstrate how Particle Swarm Optimization (PSO) can be employed to solve the Adaptive Impact Absorption (AIA) problem. We consider a truss structure which is subjected to impact loads. Stiff bars can be replaced by elastoplastic fuses which control theirs dynamical response. The point of optimization is to maximize or minimize a given objective function by redesigning the structure. This is realized by redistributing the initial mass, finding proper fuse localizations and adjusting, in real-time, the elastoplastic limits. Comparing to the previous results, we show that PSO is capable of achieving results at least as good as gradient-based optimization, having at the same time much larger flexibility regarding the definition of the objective function. This gives significantly broader field of potential applications. In particular, we present how PSO can be used to solve the simultaneous optimization problem: mass redistribution and fuse positioning for a set of expected, various impacts.

adaptive impact absorption, particle swarm optimization, truss structures

W pracy przeanalizowano dane liczbowe dotyczące wyników wyborów do sejmików wojewódzkich, które odbyły się w Polsce w listopadzie 2014. Badania korelacji pomiędzy frekwencja wyborczą, odsetkiem głosów nieważnych oraz wynikami danego komitetu wyborczego, a także porównanie częstości występowania poszczególnych cyfr w wynikach wyborczych z prawem Benforda nie wykazały oczywistych anomalii statystycznych, które mogłyby dostarczyć argumentów za tezą o masowych nieprawidłowościach lub oszustwach. Z drugiej strony porównanie wyników wyborów z lat 2010 i 2014 oraz analiza modelu matematycznego symulującego zarejestrowane wyniki pozwala potwierdzić „efekt pierwszej strony”: głosowanie za pomocą broszury wyborczej mogło przynieść wymierne korzyści komitetowi wyborczemu, który w wyborach wylosował numer 1.

wybory samorządowe, analiza statystyczna, prawo Benforda, efekt pierwszej strony

A typical unmanned and remotely operated platform is usually equipped with cameras which give insufficient information about the nearest environment and an operator has difficulties in driving such a platform in unknown environment. In this paper, a problem of the platform nearest area map building based on additional devices is considered. The platform is equipped with SICK LMS lasers, inclinometer and radars. Combining information from the devices allows to build a map which helps an operator to drive the platform more efficiently.

3D Map Building, Laser Telemeter, Navigation System, Surroundings Recognition, Unmanned Ground Platform

The paper reports initial results on modeling and simulating cooperation of rescue robots. Our aim is to investigate the efficiency of the “first simulate then act” paradigm for rescue operations undertaken by multiple mobile robots, assuming that information about the environment can be incomplete and partially inconsistent. To deal with ignorance and contradictions we adopt a non–standard logic in the inference engine. To demonstrate feasibility of our approach we present the architecture of the developed simulator and results of experiments done so far.

four-valued logic, search and rescue, multirobotic systems

The question of the origin of magnetic fields of Ap/Bp stars is still regarded as an interesting puzzle of stellar astrophysics. We investigate the possibility that the randomness and relative complexity of these fields are remnants of a magnetic instability. In the studied scenario it is assumed a priori that the surface of an Ap star is slowed down in its early evolutionary stage more than its analogous A star. This leads to a significant differential rotation in its interior, making it possible to generate a strong toroidal magnetic field in the radiative zone. Under such circumstances the kink-type Tayler instability is likely to set in. The presented numerical simulations in a compressible, spherical domain show that the instability can produce large surface magnetic fields, even of the order of 0.01 − 1 of the internal toroidal component (depending on the setup). The resulting magnetic fields can then serve as “initial conditions” evolving into a stable magnetic configuration (however, the matter of long-term stability is not addressed here). This theory naturally supports the fact that Ap stars rotate typically slower than normal A stars (the Tayler instability is suppressed when rotation is too fast), it also qualitatively explains the dependence of the apparent obliquity of the main magnetic axis on the rotation period, as well as the existence of the minimum field threshold (no Ap stars have been observed with fields weaker than ≈ 102 G). Given that the generation of the initial differential rotation and initial poloidal fields are not discussed here, the results have a speculative nature and can be viewed as a possible step toward a full understanding of Ap star magnetism.

magnetic field, chemically peculiar, magnetohydrodynamics (MHD), instabilities

We describe a system allowing a mobile robot equipped with a 3D laser range finder to navigate in the indoor and outdoor environment. A global map of the environment is constructed, and the particle filter algorithm is used in order to accurately determine the position of the robot. Based on data from the laser only, the robot is able to recognize certain classes of objects like a floor, a door, a washbasin, or a wastebasket, and places like corridors or rooms. For complex objects, the recognition process is based on the Haar feature identification. When an object is detected and identified, its position is associated with the appropriate place in the global map, making it possible to give orders to the robot with the use of semantic labels, e.g., “go to the nearest wastebasket”. The obstaclefree path is generated using a Cellular Neural Network, accounting for travel costs with distance or ground quality. This path planning method is fast and in comparison with the potential field method it does not suffer from the local minima problem. We present some results of experiments performed in a real indoor environment.

artificial intelligence, robotics, mapping

In this article we present a navigation system of a mobile robot based on parallel calculations. It is assumed that the robot is equipped with a 3D laser range scanner. The system is essentially based on a dual grid-object, where labels are attached to detected objects (such maps can be used in navigation based on semantic information). We use a classical SMPA (Sense - Model - Plan - Act) architecture for navigation, however, some steps concerning object detection, planning and localization are parallelized in order to speed up the entire process. The CUDA (Compute Unified Device Architecture) technology allows us to execute our algorithms on many processing units with use of a inexpensive graphics card which makes it possible to apply the proposed navigation system in a real time.

navigation, neural network, parallel computing

W pracy omówiono metody określenia zmian położenia robota mobilnego w środowisku 3D. Przedstawiono dwie implementacje filtru cząsteczkowego, których celem jest przyspieszenie obliczeń. Omówiono metodę, w której zmiana położenia i orientacji odbywa się oddzielnie oraz metodę wykorzystującą procesory graficzne w algorytmie lokalizacji.

Filtry cząsteczkowe, lokalizacja, mapowanie 3D

Nowadays mobile robots find application in many areas of production, public transport, security and defense, exploration of space, etc. In order to make further progress in this domain of engineering, a significant barrier has to be broken: robots must be able to understand the meaning of surrounding world. Until now, mobile robots have only perceived geometrical features of the environment. Rapid progress in sensory devices (video cameras, laser range finders, microwave radars) and sufficient computational power available on-board makes it possible to develop robot controllers that possess certain knowledge about the area of application and which are able to reason at a semantic level.

The first part of the paper deals with mobile robots dedicated to operate inside buildings. A concept of the semantic navigation based upon hypergraphs is introduced. Then it is shown how semantic information, useful for mobile robots, can be extracted from the digital documentation of a building.

In the second part of the paper we report the latest results on extracting semantic features from the raw data supplied by laser scanners. The aim of this research is to develop a system that will enable a mobile robot to operate in a building with ability to recognise and identify objects of certain classes. Data processing techniques involved in this system include a 3D-model of the environment updated on-line, rule-based and feature-based classifiers of objects, a path planner utilizing cellular networks and other advanced tools. Experiments carried out under real-life conditions validate the proposed solutions.

semantic navigation, mobile robots, laser scanners

A model of city traffic based on NagelSchreckenberg cellular automaton (CA) model is presented. Traffic control is realized at intersections with two conflicting streams each (at any time at most one stream can have „green light” assigned to it). For simple and regular lattice-like networks which are considered, it is easy to find optimal switching periods giving maximum possible flow rates. These optimal strategies are compared with a self-controlling approach proposed by Lämmer and Helbing (2008), which has not been implemented in a CA model until now. Previous work proved that generally this method gives superior results when compared to classical methods. In this paper we show that for deterministic scenario such control leads to self-organization, and that the solution always quickly converges to the optimal solution which is known in this case. Moreover, we consider also non-deterministic case, in the sense that possibility of turning with given probability is allowed. It is shown that the self-controlling strategy always gives better results than any solution based on fixed cycles with green waves.

cellular automata model, self-organizing traffic control, urban networks

W artykule przedstawiono metodę budowy hybrydowej rastrowo-obiektowej mapy otoczenia mobilnego na podstawie wskazań skanera laserowego 3D. Chmura punktów jest zapisywana w postaci zbioru wektorów normalnych. Składowe wektora są reprezentowane jako składowe RGB. Przeprowadzana jest segmentacja obrazu, a następnie dokonuje się klasyfikacji semantycznej. W procesie klasyfikacji wykorzystuje się cechy Haara oraz systemy regułowe. Każdy wykryty obiekt jest przypisywany do pewnej komórki mapy rastrowej. Mapa utworzona w ten sposób może być następnie wykorzystana w algorytmie nawigacyjnym – ułatwia współpracę robot-człowiek oraz planowanie trasy. Metoda jest kontynuacją algorytmu opisanego w artykule pt. ”Segmentacja danych otrzymanych z dalmierza laserowego 3D”.

nawigacja, reprezentacja środowiska

Otoczenie robota – wnętrze budynku jak i obszar znajdujący się na zewnątrz może być podzielony na fragmenty, którym następnie możemy przypisać pewne znaczenie semantyczne. Przed przystąpieniem do dokonywania klasyfikacji należy jednak dokonać filtracji i segmentacji danych pomiarowych. W poniższym artykule przedstawione zostaną wyniki segmentacji chmury punktów, którą otrzymujemy na podstawie wskazań laserowego skanera 3D. Zastosowano nowatorską technikę, w której dane pomiarowe zamieniane są na postać kartezjańską, następnie obliczane są wektory normalne do powierzchni, na której punkty leżą. Składowe wektora są normalizowane i zapisywane w reprezentacji RGB. W wyniku opisanej transformacji powstaje kolorowy obraz. Dzięki temu problem segmentacji danych w przestrzeni 3D jest sprowadzony do zadania analizy kolorowych obrazów. Umożliwia to zastosowanie znanych z wizji algorytmów: usuwania szumów, rozrostu ziarna i segmentacji. Przeprowadzone eksperymenty w pomieszczeniu zamkniętym i na zewnątrz budynku potwierdziły efektywność przyjętej metody.

segmentacja, mapy 3D

W poniższej pracy przedstawiono system nawigacyjny robota mobilnego. W procesie planowania trasy wykorzystuje się semantyczną wiedzę o otoczeniu. Robot wyposażony w skaner laserowy 3D analizuje otoczenie i przypisuje obserwowanym obiektom etykiety. Cel, do którego robot ma dotrzec jest wskazywany poprzez podanie nazwy obiektu. Możemy więc wydać polecenie typu jedź do ściany, do drzwi, czy też umywalki. Zastosowano hybrydową rastrowo-obiektową reprezentację otoczenia. W procesie planowania trasy zastosowano sieci komórkowe.

semantyczna nawigacja, roboty mobilne, sieci komórkowe

W artykule przedstawiono system wspomagający pracę operatora zdalnie kontrolowanego pojazdu poprzez wizualizację najbliższego otoczenia jako mapy wysokośsci przeszkód. Informacje o otoczeniu zbierane są za pomocą dalmierzy laserowych 2D dokonujących pomiaru w jednej płaszczyźnie, dane dotyczące umiejscowienia i przechyłów pojazdu dostarczają inklinometr, radarowy czujnik prędkosci oraz moduł GPS. Na podstawie przemieszczania się pojazdu możliwa jest pełna rekonstrukcja środowiska, przy założeniu że jest ono całkowicie statyczne. Przedstawiono wyniki budowania mapy 2,5D dla trzech róznych środowisk: garażu podziemnego, przejazdu między budynkami i parkingu samochodowego.

mapa otoczenia, nawigacja, dalmierze laserowe

The magnetorotational instability (MRI) is thought to play a key role in the formation of stars and black holes by sustaining the turbulence in hydrodynamically stable Keplerian accretion disks. In previous experiments the MRI was observed in a liquid metal Taylor-Couette flow at moderate Reynolds numbers by applying a helical magnetic field. The observation of this helical MRI (HMRI) was interfered with a significant Ekman pumping driven by solid end caps that confined the instability only to a part of the Taylor-Couette cell. This paper describes the observation of the HMRI in an improved Taylor-Couette setup with the Ekman pumping significantly reduced by using split end caps. The HMRI, which now spreads over the whole height of the cell, appears much sharper and in better agreement with numerical predictions. By analyzing various parameter dependencies we conclude that the observed HMRI represents a self-sustained global instability rather than a noise-sustained convective one.

magnetohydrodynamics, magnetorotational instability, Taylor-Couette flow

The magnetorotational instability (MRI) plays an essential role in the formation of stars and black holes. By destabilizing hydrodynamically stable Keplerian flows, the MRI triggers turbulence and enables outward transport of angular momentum in accretion discs. We present the results of a liquid metal Taylor–Couette experiment under the influence of helical magnetic fields that show typical features of MRI at Reynolds numbers of the order 1000 and Hartmann numbers of the order 10. Particular focus is laid on an improved experiment, in which split end caps are used to minimize the Ekman pumping.

magnetohydrodynamics, magnetorotational instability, Taylor–Couette experiment

We show that the Soft Actor-Critic Reinforced Learning algorithm is able to find efficient motion patterns for a hyper-redundant robot consisting of 6 identical modules connected in a chain like fashion. The control is done by applying relative angular velocities between the modules. Analogous system has been studied before in the context of a robotic trunk-like manipulator.

Reinforced Learning, Robotic Locomotion, Hyperreduntant Modular Manipulator, Arm-Z, Extremely Modular System

Recent research has shown that visual and inertial measurements can serve as a powerful, robust and accurate odometry source when processed by state-of-the-art algorithms. One of the main benefits of such approach is short latency, even for on-board computers working on Miniature Autonomous Vehicles (MAV). However, depending on environmental conditions or sensor motion patterns, this type of odometry may be prone to drift or even divergence. In the presented work, it is shown that employing occupancy maps can limit such undesirable behaviour while still providing pose estimate at high frequencies. This is of particular importance for highly dynamical MAV control with limited on-board numerical capabilities.

3D maps, Visual-inertial odometry, Aerial systems, MAV control, Occupancy maps

Controlling flow in networks by means of decentralized strategies have gained a lot of attention in recent years. Typical advantages of such approach – efficiency, scalability, versatility, fault tolerance – make it an interesting alternative to more traditional, global optimization. In the paper it is shown how the continuous, macroscopic, self-organizing control proposed by Lämmer and Helbing [10] can be implemented in the discrete, nondeterministic cellular automaton (CA) model of urban traffic. Using various examples, it is demonstrated that the decentralized approach outperforms the best nonresponsive solution based on fixed cycles. In order to analyse relatively large parameter space, an HPC cluster has been used to run multiple versions of a serial CA simulator. The presented model can serve as a test bed for testing other optimization methods and vehicle routing algorithms realized with the use of CA.

Urban traffic, Nondeterministic cellular automaton, Self-organizing control, Decentralized control

Truss-Z (TZ) is a skeletal system for creating free-form pedestrian ramps and ramp networks among any number of terminals in space. TZ structures are composed of four variations of a single basic unit subjected to affine transformations (mirror reflection, rotation and combination of both). This paper presents a new approach to the optimization of the layout of a singlebranch Truss-Z (STZ) in constrained environment (E). The problem is formulated as follows: create an STZ from a start (sP) to end point (eP) without self-intersections and collisions with two obstacles. This is a multi-criterial optimization problem where three independent objectives are subjected to minimization: the total number of modules (n), the “reaching error” to eP and the “overlapping error”. All three criteria are weighted and aggregated to a single cost function (CF). The calculation of CF is based on image processing of rendered geometry of individual STZs in E. The optimization is performed by population-based classic heuristic method - Evolution Strategy (ES). The computation of CF is the most time consuming, however, its parallelization is rather straightforward. Two parallelization methods are presented: distribution over Wolfram Lightweight Grid and application of general purpose graphical processing units (GPGPUs) with the use of CUDA platform.

Extremely Modular System, Truss-Z, discrete optimization, image processing, rasterization, GPU, CUDA, Mathematica, Wolfram Lightweight Grid.

Pipe-Z (PZ) is a parametric design system which comprised of a congruent modules (PZM) allows the creation of complex three-dimensional, single-branch structures which can be represented by mathematical knots. Once the geometrical parameters are set for the PZM, the shape of PZ is controlled solely by relative twists of the PZMs in a sequence. Therefore each PZM has one degree of freedom (1DOF). This paper presents the preliminary optimization of PZ reconfiguration from a “straight tube” to a half-torus. Here the displacement of PZMs transverse to the “bending direction” is to be minimized. In other words, it resembles “truing” of a wheel. In the considered case, the PZ is comprised of eight hexagonal PZMs. Thus every PZM can have six possible positions relative to the previous module. The initial (PZI) and target (PZT) configurations are given. Since the time-steps and relative twists are discrete, it is a discrete optimization and has combinatorial nature. The number of possible configurations grows astronomically with the assumed number of time-steps from one position to another and the number of PZMs. However, the optimization algorithm can be naturally parallelized. At first the concept of PZ is outlined, followed by the experiment. The results are illustrated and discussed.

Extremely Modular System, Pipe-Z, Arm-Z, discrete optimization, dihedral rotation, “snakebot”, reconfiguration.

Truss-Z (TZ) is an Extremely Modular System (EMS). Such systems allow for creation of structurally sound free-form structures, are comprised of as few types of modules as possible, and are not constrained by a regular tessellation of space. Their objective is to create spatial structures in given environments connecting given terminals without self-intersections and obstacle-intersections. In an EMS, the assembly, reconfiguration and deployment difficulty is moved towards the module, which is relatively complex and whose assembly is not intuitive. As a result, an EMS requires intensive computation for assembling its desired free-form geometrical configuration, while its advantage is the economization of construction and reconfiguration by extreme modularization and mass prefabrication. TZ is a skeletal modular system for creating free-form pedestrian ramps and ramp networks among any number of terminals in space. TZ structures are composed of four variations of a single basic module (Truss-Z module, TZM) subjected to affine transformations (mirror reflection and rotation). The previous research on TZ focused on global discrete optimization of the spatial configuration of modules. This contribution reports on the first attempts at structural optimization of the TZM for a single-branch TZ. Namely, the internal topology of a TZM and sizing of its elements are subject to optimization. An important challenge is due the fact that TZM is to be universal, i.e., it must be designed for the worst case scenario. There are four variations of each module, and due to symmetries there are thus 4^4 = 256 unique 5-unit configurations. The structural performance of all of them needs to be evaluated in terms of a typical structural criterion (the maximum von Mises effective stress), and used for structural optimization at the level of a single TZM.

Extremely Modular System, Truss-Z, Structural optimization, Effective stress

The aim of this paper is to present application of non-standard logic in the process of reasoning regarding the state of environment in a multi-robotic system. We have investigated the possibility that the usage of four-valued logic can serve as an alternative to the traditional probabilistic approach when handling the uncertainties. As a proof of concept, we have considered a typical search-and-rescue scenario in which a group of n robots performs a search for k victims localized in random positions in an initially unknown building. The building is represented in the form of a graph while the uncertainties are modeled by parameters denoting probabilities of victim observation and random blocking of a robot transition between neighboring nodes (which precludes the possibility for the robot to know anything about the actual graph structure with full confidence). The simulation results show that the unique features of applied reasoning language give significant improvement over typical approaches based on the two-valued logic. The results have a preliminary character, in the sense that up to now, only numerical experiments have been performed—applying the idea in a real multi-robotic system is currently underway.

Search-and-rescue problem, Reasoning engine, Four-valued logic, 4QL, Simulation

In this paper a system for extracting semantic information in indoor and outdoor environment from 3D laser scanner is presented. The largest objects (like walls, floor, ceiling, etc.) are recognized by constructing an RGB image based on normal vectors and applying a simple rule-based system. More sophisticated techniques are used to detect the remaining ones: Haar-like features – to classify small and irregular objects, and Cellular Neural Networks – to distinguish between different types of ground on which the robot is able to operate.

semantic information, laser range finder, navigation

Architectural design is particularly difficult, because it must combine multicriterial engineering optimization with other types of challenges, such as aesthetic and psychological which are usually ill-defined and arbitrary. Single-family house (SFH) is an archetypal architectural problem. Not only because it deals with creating of a habitat for the basic social cell, but also in a nutshell represents the entire spectrum of issues prevailing to architecture. That is environmental, both natural and artificial conditions. Since the scale of this classic problem is relatively small, it is usually manageable by a single architect. Nevertheless, the design conditions (natural and symbolical) are relatively diverse which inspire architects' imagination. This makes SFH probably the most favorable and pleasurable type of architectural design among the architects. In this multicriterial optimization certain aspects are to be minimized (e.g. noise exposition, glare, overheating, etc.) and others to be maximized (e.g. functionality, privacy, etc.). Moreover, some aspects of design have discrete nature (e.g. the functional relationships among rooms), while other are continuous (insolation, etc.). This paper presents preliminary results of the multicriterial architectural optimization of a SFH. A two-step algorithm has been applied: at first, a graph-theoretical combinatorial search produces a set of “good” architectural layouts, which are next evaluated against environmental conditions (noise, insolation and view exposition).

architectural optimization, brute-force, GPU, multicriterial