Mobile robots

Intelligent controlComputer aided decision makingFuzzy control+4

Type-2 fuzzy logic based linguistic pursuing strategy design with a real world application

The Interest in fuzzy logic has grown significantly in various working areas such as control system design, decision making, image processing, robotics, modelling and computing with words. The reason behind this huge interest is that fuzzy logic systems perform well against nonlinearities, uncertainties or unknown system dynamics since they imitate the human's way of thinking and act accordingly against not-well known or not clear enough situations. In this context, fuzzy logic systems have been implemented into various engineering applications where uncertainties and nonlinearities are high. In recent works, it has been shown that the performance of the conventional fuzzy logic systems, also known as the type-1 fuzzy logic systems, might be inadequate in defining the nonlinearities and uncertainties in some cases. Besides, it is seen that this insufficient performance of conventional fuzzy logic system moderately caused from the crisp membership grades of the implemented conventional or type-1 fuzzy sets. Thus, researchers have been proposed the type-2 fuzzy logic systems as the alternatives of type-1 fuzzy logic systems. It has been shown in several studies that type-2 fuzzy logic systems have the capability to handle or represent the uncertainties in a better way since they have an extra degree of freedom that is provided by the footprint of uncertainty. Type-2 fuzzy logic systems can be defined as the generalized versions of their type-1 counterparts. However, in handling of the type-2 fuzzy logic systems, an additional procedure is needed in type-2 fuzzy inference system that is defined as type reduction unlike their type-1 counterparts. The type reduction is applied to the antecedent type-2 membership functions of the fuzzy rules in order to reduce them into type-1 form before the defuzzification process. Type-2 fuzzy logic systems still have open problems since they have a complex internal structure and their high computational costs caused from the type reduction operation of type-2 fuzzy sets. Therefore, interval type-2 fuzzy logic systems, a special form of type-2 fuzzy logic systems, have been proposed to overcome this computational problem. Interval type-2 form of the fuzzy logic systems have becoming a dominating subject in the related fuzzy logic areas. In this context, it has been seen in several studies that especially interval type-2 fuzzy logic systems are widely used and employed since their capability to represent uncertainties in a better way and also their lower computational cost. In most of the applications of type-2 and interval type-2 fuzzy logic systems, the parameter footprint of uncertainty is usually handled as a design parameter and mostly it is tuned in order to decrease the effects of uncertainties and nonlinearities on the system performance. Therefore, there are also some studies have been performed in literature on how the size or shape of the footprint of uncertainty affects the interval type-2 fuzzy mapping. On the other hand, researchers have also worked on describing an interpretable relationship between the size of the footprint of uncertainty and the output of the interval type-2 fuzzy logic systems to use the advantages of the linguistic meaning of the interval type-2 fuzzy sets and fuzzy logic systems. In the light of these studies, interval type-2 fuzzy logic systems have become popular and widely employed to the real-world applications where unknowns, uncertainties and complexities are high such as control system design, computing with words, decision making and recently in computer and real-world games. Recently, games are gaining more and more attention among the researchers since they provide challenging, realistic and complex testbeds for computational intelligence and artificial intelligence structures and methods. In particular, pursuit-evasion game types have been commonly used in various studies as they can be defined as general problem of the many other robotic problems such as obstacle avoidance, leader-following and path-planning. The main goal of the pursuit-evasion games can be defined as; one or more pursuer tries to catch one or more evader in the probable shortest time while one or more evader tries to escape as much as possible. The various types of scenarios of the pursuit-evasion games are generally handled by the researchers who work on the area of mobile robotic. In the mobile robotic field, the pursuit-evasion games are mainly implemented to solve motion planning problems in complex, unknown and dynamic environments. In literature, the types of developed pursuit-evasion game scenarios have been shaped with respect to the limitation of the moving capabilities of the agents, number of the agents, the definition of the capture and structure of the game space. In order to solve this challenging pursuit-evasion game problem, different types of methods have been employed in literature such as dynamic programming, game theory and fuzzy logic theory. In this master thesis, we have proposed a type-2 fuzzy logic based novel approach in order to solve the pursuing problem of the real-world pursuit-evasion game; "Game of Spheros" with a new experimental setup that is constructed with spherical droids Sphero 2.0 and the BB-8. The game model has been implemented to a bounded game environment with the following specifications; the pursuer is adjusted to be faster than the evader, but at the same time to make it less maneuverable the maximum steering angle of the pursuer is restricted. In the constructed pursuit-evasion game scenario, the evader droid BB-8 is controlled by a human user while the pursuing droid Sphero 2.0 is controlled by the propose type-2 fuzzy logic based pursuing system. The proposed intelligent type-2 fuzzy logic based pursuing system is mainly composed of five components that are the vision based localization, error signal generator, control system, type-2 fuzzy strategy planner and type-1 fuzzy decision making mechanism. The vision based localization mechanism is established with a low cost webcam to localize the droids in the game environment. This feedback information is used by error signal generator to provide the necessary inputs to the type-2 fuzzy strategy planner and type-1 fuzzy decision making mechanism. Type-2 fuzzy strategy planner and type-1 fuzzy decision making mechanism are the key blocks of these intelligent pursuing system since the type-2 fuzzy strategy planner generates the reference trajectories to be followed by the pursuer and type-1 fuzzy decision making mechanism regulates the approaching behavior of the pursuer by regulating the footprint of uncertainty sizes of the type-2 fuzzy strategy planner. In the analysis of the internal structure of the type-2 fuzzy strategy planner, we have analytically derived and presented four novel type-2 fuzzy logic based pursuing strategies which are the aggressive right pursuing strategy, aggressive left pursuing strategy, smooth right pursuing strategy and smooth left pursuing strategy for the pursuit-evasion games through detailed theoretical analyses. Type-2 fuzzy strategy planner has been constructed with single input interval type-2 fuzzy logic systems and it creates pursuing trajectories to be followed by the pursuer with respect to the change of the size of the footprint of uncertainty or the height of the lower membership functions of the interval type-2 fuzzy sets. We can determine the pursuer's approaching behavior and approaching side according to the evader by regulating only one parameter besides generating pursuing trajectories with our proposed type-2 fuzzy strategy planner which can be defined as the superior part of the proposed structure. In other words, the result of this thesis study will show that it is possible to design pursuing strategies linguistically in terms of both the approaching behavior (Aggressive, Smooth) and approaching side (Left or Right) to the evader by simply tuning the footprint of uncertainty sizes of the interval type-2 fuzzy sets. We have also presented the type-1 fuzzy decision making mechanism to regulate type-2 pursuing strategy in an online manner as there is a need to change the pursuing strategy in an intelligent way. Decision making mechanism has been constructed with type-1 fuzzy logic systems based on the expert knowledge. Proposed fuzzy decision making mechanism can be also defined as a self-tuning mechanism for the type-2 fuzzy strategy planner since it regulates the footprint of uncertainty sizes of the antecedent type-2 fuzzy sets of the type-2 fuzzy strategy planner to give the chance of increasing the pursuing performance. The fuzzy decision mechanism tuned pursuing strategy is named as the intelligent pursuing strategy in this thesis. The outcomes of the thesis study will show that the superiority of the proposed intelligent pursuing strategy since fuzzy decision making mechanism provides the advantages of both aggressive and smooth pursuing strategies by regulating the footprint of uncertainty sizes of the type-2 fuzzy strategy planner in an online manner. Finally, in order to validate that the analyses on the linguistic relationships of the interval type-2 fuzzy sets of the footprint of uncertainty and also show that the performances of the proposed structure, we have performed real-world game experiments. Firstly, we will present several experimental results to validate that whether the proposed type-2 fuzzy strategy planner has an interpretable relationship between its footprint of uncertainty size and its output. Then, the pursuing performances of the proposed linguistic pursuing strategies will be also examined against a human user. Comparative experimental results of the type-2 fuzzy logic based pursuing strategies will be presented to show that resulting performances are satisfactory. It will be demonstrated that intelligent pursuing strategy, generated from the type-2 fuzzy strategy planner and fuzzy decision making mechanism structures, improved the pursuing performance as it has the capability to change the employed pursuing strategy in the dynamic pursuit-evasion game against a human controlled evader.

Aykut Beke

İstanbul Technical University · Fen Bilimleri Enstitüsü

2018

DoctorateOpen AccessEN

Quantized feedback control of autonomous robots

In this dissertation, a new robot navigation framework is introduced to close thegap between the environment mapping, path planning computations and the vehicle control problem. General approaches for robot navigation deals with the path planning and navigation with the assumption that the control system of the robot can trackthe given reference values of the planners. On the other hand, the control system architecture assumes that the incoming reference signals are suitable for its control range or commonly thresholds the out of range signals. In both situations, these systems act separately from each other, and there is no backwards information passed through these systems. As a result, even the path and motion planner outputs looks like optimal outcomes, if these outputs do not have any compliance with the control system, then the total system behavior is not accomplished as expected.To ensure connection with the planning and control structures and establish a feedback mechanism to the overall system, navigation framework employs the quantization operator both in the environment mapping and in the controller architecture. Using the quantizer in grid segmentation problem leads to a novel approach within occupancy grid literature. By using computer graphics methodologies, an efficient occupancy grid method is derived for assigning the point clouds that are gathered from 3D sensors to the grids. This assignment uses a hybrid approach, which combines the topological and grid space partitioning methods. Geometric properties of the grid voxels are interpreted as quantized spatial information from the environment and used as an input to the motion planner that generates reference signals to the controller side of the robotics system.The first main part of the navigation framework is called as input quantizer. Sensor inputs and the path splines are represented with the independent graph nodes for each axis in the 3D Euclidean space. Space partitioning methodology is executed with a binary segmentation independent for each axis until the corresponding node of the path splines and the obstacle nodes have no intersection. This binary segmentation and its reverse operation are called as zoom-in and zoom-out properties respectively in the study. After the motion planner that shapes the reference signals with using the information from the grid voxels, another main block of the framework is in charge of synthesizing an asymptotically stable controller, which is called output quantizer. As its name suggests, this structure also consists of a quantizer, which is denominated as qLQR.This method is adapted to the proposed framework to satisfy both the stability and the performance concerns.

Mobile robotsOptimal controlPath planning

Onur Şencan

İstanbul Technical University · Institute of Graduate Studies in Science

2018

Sloping groundMobile robotsMapping+2

Map planning in planes with straight and curved borders using chaotic dynamics

Map planning is the creation of a map while a robot or object is moving throughout a specified region. In this thesis, map planning is performed for planar regions in the forms of squares, rectangles, and circles. Regions containing obstacles in different forms are also considered. The rotation angle of a mobile robot in the map planning process is determined by means of chaotic solutions of systems of differential equations such as the Lorenz system, Rössler system, Duffing oscillator, and a hyperchaotic system. The rotation angle is chosen as one of the coordinates of the chaotic solutions. Region covering speeds for different chaotic systems are compared. To the best of our knowledge, this is the first time in the literature that map planning is performed for planar regions with curved borders.

Ömer Faruk Tekin

TED University · Lisansüstü Programlar Enstitüsü

Computer visionMobile robotsMachine vision

DoctorateOpen AccessEN

Place recognition with an attentive robot

This thesis is concerned with robotic place recognition. From the robot's perspective, the goal of place recognition is to determine its whereabouts - given its knowledge of places it has visited previously. This problem is broken into four stages: First, the problem of how the robot should look around in the current place is considered and a model of attentive visual behaviours based on a family of artificial potential functions is developed so that the robot is able to consider the interesting points in the scene. Next, the problem of place representation is considered. For this, a topological model referred to as bubble space is developed. Sensory features associated with all the interesting points are encoded as bubble surfaces - in a manner that is implicitly dependent on robot pose. For compact representation, bubble surfaces are transformed into bubble descriptors that are rotationally invariant with respect to heading changes while being computable in an incremental manner as each new set of visual observations is made. Bubble descriptors are used for supervised place learning and recognition with support vector machines. Following, the problem of integration of sensory and semantic cues is considered and a representation referred to as ``cue descriptor semantic forest'' is proposed. The integration of different feature and semantic cues becomes very simple regardless of their type or their number while allowing relative and possibly partial semantic expressiveness via cue weights. Finally, the problem of unsupervised place learning is considered. A novel approach for the incremental and autonomous construction of bubble descriptor semantic trees is proposed. Comparative experiments on benchmark datasets as well as real-time robotic experiments indicate high recognition rates, dynamic with low computation times.

Özgür Erkent

2013

ExerciseMotion sensorsMobile robots+2

Socially assistive child-robot interaction in physical exercise coaching

The main contribution of this thesis is the design and implementation of an autonomous socially assistive human robot interaction system to engage children in performing several physical exercise motions by providing real-time feedback, guidance, and encouragement. Preliminary offline tests were done within a rehabilitation therapy context but since the potential of the system to encourage such exercise activities is beyond the physical therapy fields, we designed an online system to behave as an exercise coach that can motivate children to start and continue improving their health and become physically active. The online system was tested in several preliminary experiments with an exercise coach and children during the design process. Furthermore, an experimental study was conducted with 19 healthy children aged between 4 and 12 to test the feasibility and the effectiveness of our online Socially Assistive Robotic (SAR) exercise system across a variety of performance and evaluation measures. The results of the study validate the effectiveness of the system in motivating and helping children to complete physical exercises. The children engaged in physical ex- ercise throughout the interaction sessions with a robot coach and rated the interaction highly in terms of enjoyableness, and rated the robot exercise coach highly in terms of social attraction, social presence, and companionship via a questionnaire answered after each session.

Arzu Güneysu

Smart agricultureMobile robotsModel predictive control+1

Cooperative localization and control for heterogeneous mobile robots

This thesis contributes to the UTOPIA project, which focuses on developing cooperative localization and control strategies for heterogeneous mobile robots in agriculture. The research involves designing an Unmanned Aerial Vehicle (UAV) and an Unmanned Ground Vehicle (UGV), each with distinct sensors and maneuvering capabilities. These differences require a robust localization strategy for effective cooperation. The Extended Kalman Filter (EKF) was tested in various configurations to improve localization accuracy. Enhancements were first achieved through simulations and later validated in experiments. The EKF's performance was assessed in three cooperative missions, showing that integrating more sensor data into the EKF significantly reduced the Root Mean Square Error (RMSE), improving localization accuracy. While homogeneous robot systems benefit from consistent localization with similar sensors, heterogeneous systems face challenges due to varying sensor capabilities. This research underscores the importance of robust localization for effective cooperation in diverse robotic systems. UAVs, despite their fast movement and wide vision, are limited by load capacity and battery life, whereas UGVs offer greater load capacity and battery longevity but move slower. These findings highlight the need for effective cooperative localization and control strategies in complex environments.

Ahmet Mustafa Kangal

2024

Active learningMobile robots

Unsupervised, continual and active object learning by mobile robots

This thesis is concerned with autonomous object cognition in mobile robots. A key issue in open-world environments is that new objects emerge frequently, and hence labeled data are unavailable. These conditions call for unsupervised and continual learning methods that enable autonomous object cognition. Moreover, their inherent mobility enables active learning, allowing robots to explore and gather diverse sensory information to improve object understanding. This thesis proposes an approach that integrates unsupervised, incremental and active learning to address these challenges, enabling mobile robots to autonomously acquire, refine, and maintain object representations in open-world environments. We assume that the robot is able to segment each incoming RGB-D frame, track the resulting segments across the frames, and accumulate the segments associated with each tracked object. We also assume that the robot is able to navigate as it finds necessary to do so. First, object representation is considered, and an extensive evaluation of the existing models is conducted to identify those achieving the best performance. Following, existing one-class classifiers are evaluated with respect to learning and recognition performance. Thirdly, unsupervised and incremental object learning is addressed via an approach in which an object memory is developed. Finally, active object learning is considered through having the robot be selective with regards to the data it collects in order to learn new objects. All the proposed methods are tested extensively in both the Gazebo simulation environment and real-world scenarios with a mobile robot. This work contributes to the advancement of open-world robotic perception by enabling adaptive, label-free object understanding in new environments.

Haluk Ziya Zorluoğlu

2026

Dinamik kontrolGerçek zamanlı kinematikHareketli robotlar+2

İnsansı robotların tüm vücut kinematik ve dinamik modellenmesi ve kontrolü

Yetmişli yılların başlarından itibaren insansı robotların mühendislik alanındaki araştırmaları artan bir ilgiyle devam etmektedir. Temel olarak insansı robotların insanların günlük yaşamlarına dahil olup yaşam kalitelerini arttırmaları beklenmektedir. İnsansı robotların talep edilmesindeki en önemli unsur mekanik yapılarından kaynaklı olan yüksek hareket kabiliyeti kapasiteleridir. İnsansı robotlar insanların günlük yaşam uygulamalarını gerçekleştirmede kullanılabilecek en uygun mekanizmalardan bir tanesi olarak kabul edilmektedir. İnsansı robotlar mekanik yapılarından dolayı oldukça yüksek hareket kabiliyetine sahip olsalarda mevcut insansı robotlar bu kabiliyetin çok az bir kısmını kullanabilmektedirler. İnsan hareketlerinin iyi bir şekilde anlaşılamamış olması ve basitleştirilmiş model ve kontrolörlerin kullanılması insansı robotların hareket yeteneklerini kısıtlayan başlıca etmenlerdir. Bu tezde insansı robotların hareket yeteneklerinin geliştirilmesi hedeflenmişitir. Bunun için insansı robotların tüm vücut dinamik modeli elde edilmiş ve insanların günlük yaşamlarında yaptıkları davranışları gerçekleştirebilmelerini sağlayacak gelişmiş kontrolörler tasarlanmıştır. Bu tez 6 bölümden oluşmaktadır. Her bir bölümün içeriği aşağıda açıklanmaktadır. İlk bölümde tez ile ilgili temel kavramlar açıklanmış, tezin amacı ve motivasyon kaynakları belirtilmiş, literatür özeti verilmiş, problemin tanımı yapılarak önerilen çözümler açıklanmıştır. İkinci bölümde insansı robotların kontrolünde kullanılan kontrolörler açıklanmıştır. Bu tezde bozucu gözetleyici tabanlı kontrolörler dayanıklı kontrolün ve sensörsüz kuvvet kontrolünün elde edilmeleri için kullanılmıştır. Üçüncü bölümde insansı robotların tüm vücut dinamik modeli elde edilmiştir. Tüm vücut dinamik model elde edilirken kayan nokta dinamiği kullanılmıştır. Geleneksel dinamik modellere kıyasla kayan nokta dinamiği tabanlı yaklaşımın temel avantajı insansı robotlar için sürekli yapıya sahip bir dinamik modelin elde edilebilmesine olanak sağlamasıdır. Bu bölümde dinamik model nümerik ve analitik tabanlı iki farklı yöntem kullanılarak elde edilmiştir. Elde edlen dinamik model ve MATLAB'ın sanal gerçeklik araç kutusu (Virtual Reality Toolbox) kullanılılarak yeni bir simülatör tasarlanmıştır. Dördüncü bölümde insansı robotların alt vücut kontrolü üzerinde durulmuştur. Alt vücut kontrol uygulamalarında iki bacaklı yürüyen bir mekanizma kullanılmıştır. İki bacaklı robot statik ve dinamik yürüme algoritmaları kullanılarak kontrol edilmiş ve tasarlanan simülatör kullanılarak benzetim sonuçları sunulmuştur. Ayrıca iki bacaklı robotun engebeli bir yüzeyde dinamik yürüme kontrolü gerçekleştirilmiştir. Beşinci bölümde insansı robotların tüm vücut kontrolü üzerinde durulmuştur. Üst vücut kontrolü serbest ve kısıtlı hareket kontrolleri olmak üzere iki temel bölümde incelenmiştir. Serbest hareket kontrolünde robot kolları çevre ile etkileşime girmezken kısıtlı hareket kontrolünde robot kolları bir nesne ile temas halindedir. Serbest hareket kontrolünde robot kolları için bağımsız yörüngeler tasarlanmıştır. Robot kollarının yörüngeleri kullanılarak dinamik yürümenin kararlılığı iyileştirilmiştir. Kısıtlı hareket kontrolünde robotun bilinmeyen bir nesne ile etkileşime girmesi incelenmiştir. Güvenli kontak kontrolü ve bilinmeyen bir nesnenin ittirilmesi ile ilgili simülasyon çalışmaları gerçekleştirilmiştir. İnsansı robotların hareket kabiliyetlerini insanların hareket yeteneklerini kullanarak arttırmak için çift taraflı kontrol yapısı önerilmiştir. Bir çok farklı uygulamada kullanılabilecek iki farklı çift taraflı kontrol yapısı sunulmuştur. Altıncı bölümde sonuçlar ve öneriler verilmektedir.

Emre Sarıyıldız

İstanbul Technical University · Fen Bilimleri Enstitüsü

Active interactive learningLanguage gamesEducation plays+4

Robot and avatar based interactive system for children's education

The aim of this thesis is to present usable educational tools of students with special needs. Hearing impaired students cannot use text and speech based technological, educational material that is becoming a crucial tool for modern education. While they are not a direct substitute for human teachers, technological support materials in education have been shown to motivate children and aid instructors. To allow the use of such methods, three applications aimed at educating deaf children using the robot, the web, and avatar-based technologies was designed. One such technological medium that can be used to connect with such children are robots. Robots are often used in children's education by making use of their inherent personified nature compared to other non-human teaching mediums. Children prefer to interact by touching during a training session. In this study, we developed a method, where two humanoid robots namely R3 and NAO are used to perform sign language gestures to communicate with children using sign languages with creating human-like hand, arm and body gestures. The most powerful aspect of the robot is its embodiment that causes children to interact with the robot socially, rather than treating it as a simple, book like education tool. One other educational tool we have developed is a signing avatar. Like videos of signers, a signing avatar for TİD would allow communicating information in the form of visual gestures, thus becoming usable when the use of text is infeasible. However, compared to videos, animated avatars offer several advantages like easy reproducibility of gesture sequences, control of point of view, adjusting the speed of the sign and being smaller in storage and bandwidth then videos. For this thesis, a signing avatar was developed to represent a portion of the social sciences course for Turkish primary education curriculum. Finally, a web and android based experimental framework was developed to perform customized human-machine interaction experiments. Sign language and drum based interaction game schemes were developed to perform in-depth analysis of different education based modalities with users from different demographics. By performing usability studies and evaluations in a gamification, scheme using turn-taking games, the effectiveness of the described three methods were analyzed. The observations and the reactions of children demonstrate that of the three technologies implemented, robot technologies may be the most effective method for assisting children. While a prototype interaction scheme and used in controlled education settings have been constructed, in their current technology level, price range and ease of operability robots such as Nao do not present an implementable option of presenting the entire primary school corpora. For that reason, a signing avatar and recreated content from the Social Sciences corpora have been implemented. While the signing avatar an imitation of videos created by signing users, it possesses the advantage of easily synthesizing novel sequences of content from existing isolated sign videos. To analyze the success and effectiveness of the developed methods, the performance of the text, signer video and avatar-based interaction in a human-computer interaction scheme using elementary school students have been compared. The results demonstrate that of the three methods, video-based methods were the most comprehensible closely followed by avatar based methods among kids who did not know how to read. Since the aim was to make the signing avatar as comprehensible as possible, the results demonstrate that goal have been.

Rabia Yorgancı Kındıroğlu

İstanbul Technical University · Fen Bilimleri Enstitüsü

Distributed computer systemImage processing-computer assistedMobile robots+1

A distributed human identification system for indoor environments

Nowadays "internet of things" and "robotics" topics are getting more and more popular because "state of the art" technologies are easier to be applied in our daily lives. Utilization of "internet of things" started home automation systems to use devices at home remotely. While these automation systems aim at giving access to designated areas, they also offer solutions for unauthorized access problems by keeping logs of video records and control mechanisms as well as features for more security measures. Besides, developers are able to implement and monitor video surveillance systems to support security in the IoT platforms using low cost embedded computers. This study presents an entire system, which is similar to home automation system and allows the login of a person entering a private property using face recognition system, video calls, sending notifications and having more components such as threat detection. In this study statically positioned camera systems were implemented to identify people thanks to face recognition algorithms. Captured images from fixed positioned camera systems(FPFRS) had problems to process on identification. The problems were differences in distance of people to the camera and differences caused by posing. Therefore a mobile robot that has camera, was incorporated to recognize faces of people in the environment. The mobile robot can track people and keep the distance to the people constant. Thus, the mobile robot can solve the distance problem. However, the mobile robot can not track people perfectly without miss. Sometimes it can have trouble while tracking people. Thus, integrating the mobile robot to the Fixed Positioned Face Recognition System has a big potential to deal with the face recognition problem in indoor environments. Therefore an distributed approach was proposed to recognize humans in this study. The system utilizes 1) Cameras statically mounted on wall devices (FPFRS), 2) Cameras mounted on mobile robots. FPFRS were built on a tiny embedded computer. Motion sensor and infrared camera components were used to gather data about environment. However, sometimes they can fail to recognize faces because of their fixed positions. This decreases face recognition performance of the system when FPFRS tries to recognize person from captured images. Because distance between camera and person is very important in order to obtain features of face. To improve the performance of this system proposed method given in this thesis aims at solving the problem of fixed position with robots. Therefore a mobile robot is added to improve face recognition accuracy gathering face images from certain distance and moving around face image that is aligned to the camera of robot before captured. With the help of the mobile robot, the system becomes more interactive and information gathering from environment process is done more efficiently. As a result proposed hybrid system in this study not only uses fixed positioned captured images by FPFRS but also uses images that are taken from mobile robot. To provide network communication between FPFRS and the mobile robot which are linked to in this distributed system, modified techniques are used which follow the rules of client-server approach. Each of the FPFRS and the mobile robot runs main program during scenarios. Inputs can be taken from the environment and enable the trigger mechanism. An FPFRS schedules its jobs and behaves according to the active jobs. After the job is done, FPFRS deactivates the jobs in the database records. These records can be manipulated from other devices. Thus communication and trigger mechanism between devices is established. These systems are more complex and it is hard to follow this kind of systems, actions. Architecture is designed to expand by adding new FPFRSs or mobile robots to the system and each FPFRS can communicate via server. One FPFRS and a mobile robot have scenario to capture more images of people from different positions. Then each device runs face recognizer and obtains a result which consists of name of person and confidence of face recognition score. Using the results of face recognizer, various techniques, including machine learning algorithms, are applied to improve accuracy of face recognition. In addition to choosing prediction based on highest confidence in face recognition, random forest tree, decision tree, and linear SVM methods are applied to address the issue of face recognizer improvement. Described systems are implemented and tested systematically based on various scenarios. To show the impact of the effect of changes on accuracy of face recognition system, experiments were done with and without the mobile robot. The proposed system was evaluated using 1)Face recognition scores on a face database, 2)Performance of video calls, and 3)Case scenarios performed on human test subjects. We have verified the effectivity of our approach by using six scenarios. In the first scenario button triggered human identification has been evaluated. In the second scenario motion triggered human identification has been analyzed. In the third scenario human identification using rotational tracking has been investigated. In the fourth scenario human identification using displacement tracking has been examined. In the fifth scenario human identification using displacement and voice combined tracking has been illustrated. In the sixth scenario human identification using robot initiated dialog has been analyzed. One or more FPFRS and the mobile robot have scenario to route the mobile robot to capture images of people on designated area of motion sensor triggered FPFRS based on the scenarios given above.

Emre Sercan Aslan

İstanbul Technical University · Fen Bilimleri Enstitüsü

Mobile robotsTarget localization

Combination of MCL and ICP methods for accurate indoor localization

Nowadays, mobile robots that are widely used in the industry are moving on pre-planned paths. These paths are prepared using wires, magnetic tapes or any other methods that allow the robot to follow the path. However, using these paths restricts the movement of the robot. This is because when it is necessary to change the target locations of the robot in the factory, the paths have to be prepared from the very beginning in accordance with the new task of the robot. Doing this will cause extra cost and time loss. For these reasons, mobile robots used in the industry need to reach the target without any help, instead of moving on predetermined paths as they go to the target position. In order to do this, they need to know their current pose. In this thesis, the localization study for indoor mobile robots that are used in the industry was performed in the simulation environment by using a two-step indoor localization method. In the first step of the localization method, the pose of the robot was roughly estimated and then its pose was roughly tracked in a known environment by using the Monte Carlo Localization algorithm. In the second step, when the robot was close to a predetermined target point, the instantaneous pose of the robot was accurately estimated by using a variant of ICP algorithm.

Erim Vural

İstanbul Technical University · Institute of Graduate Studies in Science

Endüstriyel robotlarHareketli robotlarSığır gübresi+1

Master'sOpen AccessTR

Zemin dezenfeksiyon robotu tasarım ve optimizasyonu

Hayvancılıkta robotlaşmanın amacı, çiftlik işlerini ve hayvan yönetimini daha verimli, etkili ve sürdürülebilir hale getirmektir. Bu amaçla çeşitli robotik teknolojiler ve otomasyon sistemleri kullanılarak çiftlik operasyonları optimize edilir. Robotlar, tekrarlayan ve zaman alıcı işleri daha hızlı ve hassas bir şekilde gerçekleştirerek iş gücü verimliliğini artırır. Bu, işletmelerin daha fazla üretim yapmasına ve maliyetleri düşürmesine yardımcı olur. Hayvancılık işleri sıkça zorlu koşullar gerektirir. Robotlar, insanların bu zorlu koşullara maruz kalmadan işleri yapmasını sağlar, böylece işçi sağlığı ve güvenliği iyileştirilir. Robotlar gece-gündüz, hafta içi veya hafta sonu demeden sürekli olarak çalışabilirler. Bu, işletmelerin kesintisiz üretim yapmasını sağlar. Daha iyi besleme yönetimi ve atıkların etkili bir şekilde yönetilmesi gibi robotik çözümler, çevresel etkiyi azaltabilir ve sürdürülebilirliği artırabilir. Robotlar, yapay zeka, görüntü işleme ve otomasyon gibi teknolojileri kullanarak, hayvan sağlığı, besleme yönetimi gibi konularda uzmanlık sağlayabilirler. Sonuç olarak tezimizin amacı, daha etkili, verimli, sürdürülebilir ve insan odaklı robotik hayvancılığı destekleyerek çiftlik sahiplerinin yükünü azaltırken üretim gücünü arttırıp sürdürülebilirlik sağlamak.

Ali Munğan

Pamukkale University · Fen Bilimleri Enstitüsü

2025

Mobile robotsMachine learning

Transfer learning for continuous control

Agents trained with deep reinforcement learning algorithms are capable of performing highly complex tasks including locomotion in continuous environments. In order to attain a human-level performance, the next step of research should be to investigate the ability to transfer the learning acquired in one task to unknown tasks. Concerns on generalization and overfitting in deep reinforcement learning are not usually addressed in current transfer learning research. This issue results in simplistic benchmarks and inaccurate algorithm comparisons due to rudimentary assessments. In this thesis, we propose novel regularization techniques exclusive to policy gradient algorithms for continuous control through the application of sample elimination and early stopping. By discarding samples that lead to overfitting via strict clipping we will generate robust policies for a humanoid with high generalization capacity. We also suggest the inclusion of training iteration to the hyperparameters in deep transfer learning problems. We recommend resorting to earlier snapshots of parameters depending on the target task due to the occurrence of overfitting to the source task. We demonstrate that a humanoid is capable of performing forward locomotion in unseen environments with different gravities and tangential frictions using strict clipping and early stopping. Furthermore, we evaluate our propositions on a delivery task where a humanoid is required to carry a heavy box while walking and inter-robot transfer tasks where the humanoid transfers its learning to taller and shorter robots. Because source task performance is not indicative of the generalization capacity of the algorithm we propose three different transfer learning evaluation methods. We increase the generalization capacity of a state-of-art adversarial algorithm by introducing entropy bonus, proposing different critic architectures and using simpler adversaries. Finally, we evaluate the robustness of these adversarial algorithms on morphologically modified hopper environments and environments with unknown gravities according to the criteria we proposed.

Suzan Ece Ada

Boğaziçi University · Institute of Graduate Studies in Science

2019

Boğaziçi University · Institute of Graduate Studies in Science

A reinforcement learning based controller to minimize forces on the crutches of a lower-limb exoskeleton

The majority of the metabolic energy consumption of a lower-limb exoskeleton user comes from the upper body effort, since the lower body can be considered to be passive. However, the upper body effort of lower limb exoskeleton users is ignored during motion controller development process in the literature. In this thesis study, deep reinforcement learning is used to develop a locomotion controller that minimizes the ground reaction forces (GRF) on crutches. The rationale for minimizing the ground reaction forces is to minimize the upper body effort of the user. A model of the human-exoskeleton system with crutches is created in URDF and XML formats. Reward functions that encourage the forward displacement of the center of mass of the exoskeleton-human system without falling and extreme joint torques are shaped. The state-of-the-art methods, Twin Delayed Deep Deterministic Policy Gradient (TD3) and Proximal Policy Optimization (PPO), are employed with the RaiSim and MuJoCo physics simulators and with different algorithm specific parameters in multiple training trials. The employed networks generate the joint torques based on the joint angle and velocities along with the ground reaction forces on feet and crutch tips. These generated joint torques are directly sent to the exoskeleton model and a new state is observed after implementing the action that the deep RL framework provides. Policies trained by the TD3 and PPO methods on RaiSim are observed to fail to generate proper control commands for a stable and natural looking gait. In general, it is observed that the PPO method generated higher rewards than the TD3 method on RaiSim. After failing to develop a desired policy with RaiSim, MuJoCo is employed as the simulator. Eventually, a policy that can generate a reasonable gait with a desired crutch usage and with 35% minimization in GRFs with respect to the baseline policy is developed.

Mobile robots

Aydın Emre Utku

Nonlinear control theoryNon-linear optimizationMobile robots+1

Modular safety-critical control of legged robots

With recent performance improvements, legged robots will soon enter our lives to stay. Various control algorithms are already employed in deploying existing robots, and many more algorithms are still in the making. Safety concerns during the operation of legged robots must be addressed to enable their widespread use and ease their development. Especially machine learning-based control methods would benefit from model-based constraints to improve their safety. This thesis presents a modular safety filter to improve safety, i.e., reduce the chance of a fall of a legged robot. The availability of a robot capable of locomotion is assumed, i.e., a nominal controller exists. During locomotion, terrain properties around the robot are estimated through machine learning which uses a minimal set of proprioceptive signals. A novel deep-learning model utilizing an efficient transformer architecture is used for terrain estimation. A quadratic program combines the terrain estimations with inverse dynamics and a novel control barrier function constraint to filter and certify nominal control signals. The result is an optimal controller that acts as a filter and the filtered control signal allows the safe locomotion of the robot. The resulting approach is general and could be transferred with low effort to any other legged system.

Berk Tosun

Industrial robotsMobile robots

DoctorateOpen AccessEN

Development of a mobile robot performing transport implementations in a manufacturing plant

daha sonra doldurulacaktırThis thesis investigates the development of a mobile robot engineered to enhance material transportation processes and improve efficiency within contemporary manufacturing facilities. The primary aim is to address the inefficiencies and errors inherent in manual material handling by implementing advanced robotics and automation technologies. The research encompasses several comprehensive stages: mechanical design, sensor technology integration, and software development, all coordinated through the Robot Operating System (ROS) framework. The thesis commenced with defining the robot's specifications based on the operational requirements of the manufacturing environment. This was followed by the creation of a detailed mechanical design using 3D parametric modeling to ensure the robot met the necessary specifications for payload capacity, maneuverability, and adaptability to factory floor conditions. The design phase included mechanical static structural analysis and topology optimization, ensuring structural integrity with total deformation maintained below 0.2 mm under maximum load conditions. A skid steering movement system was chosen for its reliability and efficiency in navigation. In the electronic integration phase, various advanced sensors were selected to provide essential data acquisition and control capabilities for autonomous operation. Key components included LIDAR for precise navigation, IMU sensors for orientation, encoders for movement tracking, and stereo cameras for enhanced environmental perception. These sensors were integrated with microcontrollers and an onboard computer, facilitating the necessary processing power for the robot's tasks. Wireless communication between the microprocessor and microcontroller was also established to ensure seamless operation. The software development phase involved creating robust algorithms for path planning, obstacle avoidance, and task execution. The ROS framework was utilized to facilitate communication between hardware components and develop control algorithms. Specific tools within ROS, such as Gmapping for mapping, Adaptive Monte Carlo Localization (AMCL) for positioning, and a Hybrid A* algorithm for path planning, were employed. An Extended Kalman Filter (EKF) was used to obtain accurate odometry data, essential for precise localization and navigation. Extensive simulations were conducted to refine these algorithms before real-world implementation. The final prototype underwent rigorous testing in both virtual and live production environments. Key performance metrics, such as carrying capacity, speed, accuracy, and reliability, were evaluated. The robot achieved an average transport speed of 0.5 m/s and a positioning accuracy within 5 cm, demonstrating its capability to navigate and avoid obstacles autonomously. It handled an average payload of 100 kg with a maximum deviation of less than 2% in path accuracy. Additionally, the robot's battery life supported continuous operation for up to 8 hours, proving its suitability for long shifts in manufacturing settings. The results of this study demonstrate that the developed mobile robot significantly improves operational efficiency and accuracy in material transport within manufacturing plants. Its ability to autonomously navigate and execute transport tasks with minimal human intervention represents a substantial advancement in industrial automation. This thesis not only highlights the robot's potential to optimize material handling processes but also provides groundbreaking insights into the practical application of robotics and automation technologies. The findings serve as a vital resource for researchers and industry professionals dedicated to enhancing industrial process efficiency and embracing the future of automated manufacturing.

Neslihan Demir

Aydın Adnan Menderes University · Fen Bilimleri Enstitüsü

2024

İnönü University · Fen Bilimleri Enstitüsü

Çoklu robot sistemleri için blokzincir tabanlı dağıtık görev paylaşım sistem mimarisinin geliştirilmesi

Çoklu robot uygulamalarında; robot yönetimi, robotlara görev atama, görev takibi, robotlar arası iletişim ve koordinasyon, veri depolama gibi operasyonel süreçler merkezi bir sunucu sistemi üzerinden yürütülmekte ve en kritik kararlar robotlara bırakılmadan bu sunucu tarafından alınmaktadır. Ancak merkezi sistemlerin sunucu bakım zorunluluğu, tüm iletişimin merkezi birim üzerinden yönlendirilmesi, kararların bir merkezden verilmesi, tek arıza noktası riski, olası iletişim sorunlarının (özellikle açık arazideki çoklu robot sistemlerinde karşılaşılan büyük haberleşme problemleri) bütün sistemin işleyişini etkilemesi, robot sayısı arttıkça bant genişliği gereksinimlerinin artması gibi birçok dezavantajı vardır. Bunları aşmak adına, bu tez çalışmasında robotlar arası iletişim için blokzincir tabanlı dağıtık ve merkeziyetsiz bir sistem geliştirilerek bu sistemde çalışan robotlar arasında görev paylaşımı uygulaması gerçekleştirilmiştir. Böylece robotların merkezi sunucu kullanmaksızın bir blokzincir ağı üzerinden doğrudan iletişim kurabildiği dinamik ve dağıtık bir mimari tanıtılmış ve uygulanmıştır. Blokzincir platformu olarak avantajları nedeniyle IBM tarafından geliştirilen Hyperledger Fabric (HLF) kullanılmıştır. Sistemdeki her robot HLF ağında ayrı birer fiziki üye olarak tanımlanmıştır. Merkezi sistemlerde kolaylıkla yapılabilen görev paylaşımını, dağıtık - merkeziyetsiz sistemlerde "ortak karar" niteliğinde optimum düzeyde yapmak çok büyük bir problemdir. Tez çalışmasında geliştirilen mimari ile robotlar bir "maliyet fonksiyonu" kullanarak kendileri için en uygun maliyetli görevleri seçerken; aynı zamanda ağdaki iletişim yükü de optimum seviyede tutulmaktadır. HLF ağındaki üye robotlar görev paylaşımı konusunda uzlaştıktan sonra görev verileri soket bağlantısı üzerinden robotlarda koşturulan ROS'a iletilir. Ağdaki görev paylaşım süreci; yalnızca o esnada "meşgul olmayan" robotların dahil edileceği şekilde yapılandırılmış, bu sayede "meşgul" robotların süreci etkilemesi önlenmiştir. Gerçek robotlar ve simülasyonda yapılan deneylerde önerilen sistemin tüm koşullar için optimum görev tahsisini gerçekleştirdiği görülmüştür. Bu tez çalışmasının literatüre yenilikçi katkılar sunarak, merkezi yaklaşıma güçlü bir alternatif oluşturulmasında ilham verebileceği öngörülmektedir. Geliştirilen sistem, tüm otonom sistemlere uyarlanabilir olup; hem endüstriyel hem de askeri uygulamalarda yüksek potansiyele sahiptir.

Hareketli robotlar

Mehmed Oğuz Şen

2025

Hareketli robotlarMakine öğrenmesiMakine öğrenmesi yöntemleri+2

Mobil bir robotun yapay zeka destekli kendinden öğrenebilir sürüş algoritmasının geliştirilmesi

Mobil robotlar, günümüzde otomasyon ve yapay zeka teknolojilerinin giderek artan bir şekilde entegre edildiği önemli bir alan haline gelmiştir. Bu robotlar, hareket kabiliyetine sahip olmaları sayesinde çeşitli uygulama alanlarında kullanılmaktadır. Bu uygulama alanları arasında lojistik, tarım, inşaat, sağlık ve endüstriyel otomasyon gibi birçok sektör yer almaktadır. Mobil robotların etkin ve güvenli bir şekilde hareket etmelerini sağlayan temel bileşenlerden biri sürüş algoritmalarıdır. Geleneksel sürüş algoritmaları, robotların belirli durumlar için önceden programlanmış talimatlarla hareket etmelerini sağlar. Ancak, bu yaklaşım basit ve tekrar eden görevler için etkili olabilirken, karmaşık ve değişken ortam şartlarında robotların başarılı olmasını sınırlamaktadır. Yapay zeka teknolojilerinin robotlara entegrasyonu, mobil robotların çevrelerini algılamalarını ve deneyim tabanlı öğrenme sağlayarak sürüş algoritmalarını daha adaptif ve otonom hale getirmeyi amaçlar. Bu çalışmada, mobil bir robotun yapay zeka destekli kendinden öğrenebilir sürüş algoritmasının geliştirilmesi hedeflenmiştir. Algoritmanın adaptif öğrenme mekanizmaları, uyumluluk ve otonomi, öğrenme hızı ve performansı ile gerçek dünya uygulamaları açısından literatüre değerli katkılar sunması beklenmektedir. Bu amaçla çalışmada literatürdeki yaygın olarak kullanılan makine öğrenimi metodlarından olan pekiştirmeli öğrenme (DDPG, TD3, PPO, SAC, A2C) algoritmalarından faydalanılarak karmaşık ortamlardaki başarıları gözlemlenmiş ve karşılaştırılmıştır. Oluşturulan karmaşık ortamlar her sahnede değişecek şekilde hazırlanarak tek bir algoritmanın farklı karmaşık düzeylerine sahip ortamlardaki başarıları eğitilerek test edilmiştir. Elde edilen bulgulara göre TD3 %86.3, A2C %89,2 SAC %93.6, PPO %97.6 başarı sağlamıştır. Bu sonuçlar, mobil robot teknolojisinin gelecekteki gelişimi ve çeşitli sektörlerde etkin bir şekilde kullanılması için önemli bir adım olarak görülmektedir. Sonuç olarak, bu tez çalışması mobil robotların farklı senaryolarda güvenilir ve otonom bir şekilde çalışmasını sağlayarak çeşitli uygulama alanlarında kullanılabilirliğini artırmayı hedeflemektedir. Algoritmaların çevresel değişikliklere hızlı adaptasyon yeteneği sayesinde, robotların daha genel amaçlı ve esnek bir şekilde çalışması sağlanmıştır. Dinamik ortamlar ve kapsamlı analizlerle desteklenen bu yaklaşım ile, mobil robotların otonom hareket kabiliyetlerini geliştirirken, gelecekte karmaşık ortamlarda test edilmesi ve farklı veri kaynakları ile entegrasyonlarının sağlanması öngörülmektedir. Bu çalışmanın, otonom sistemler ve yapay zeka tabanlı robotik uygulamalar için önemli bir katkı sunacağı değerlendirilmektedir.

Taner Yılmaz

Fırat University · Institute of Graduate Studies in Science

2025

Endüstriyel robotlarFPGAFırçasız motor sürücüleri+3

Yerden yüksekliği değiştirilebilen eklemli tekerlek yapısına sahip bir güvenli taşıma aracının tasarımı ve FPGA ile kesir dereceli kontrolü

Tezin konusu olan Güvenli Taşıma Aracı (GTA) ile ilgili literatürde çeşitli çalışmalar mevcuttur. Ancak hareket kabiliyetinin ve performansının arttırılması amacı ile önerilen eklem bacaklı tekerlek yapısına sahip bir GTA'nın çalışır halde bir örneği bulunmamaktadır. Benzer yapıda bir araç tasarımı ise henüz başlangıç aşamasındadır. Literatürdeki çalışmalardan farklı olarak yönünü değiştirmeden belirli yükseklikteki engellerin üzerinden ilerleyebilen ve eklem açıları ayarlanarak yerden yüksekliği değiştirilerek farklı yüksekliklerdeki yükleri alıp bırakabilen bir tasarım yapısı, tez çalışmasında önerilen robotun en önemli özgün değeridir. İleri robotik teknolojilerinin günümüz örneklerinde sıklıkla kullanılmaya başlanan Fırçasız Doğru Akım Motoru (FDAM) bu çalışmada tahrik elemanı olarak kullanılmıştır. FDAM'ların yüksek performanslı olarak kontrolü için modern kontrol yöntemlerinden olan Kesir Dereceli Kontrol (KDK) yöntemi kullanılmış olup, KDK algoritmaları FPGA (Field-Programmable Gate Array - Alanda Programlanabilir Kapı Dizisi) kartları üzerinde gerçek zamanlı olarak gerçekleştirilmiştir. Henüz literatürde FDAM'ın KDK'sına ilişkin FPGA ortamında başarılı bir gerçek zamanlı uygulama mevcut değildir. Çok sayıda FDAM'ın aynı anda ve yüksek doğrulukla kontrolünün gerçek zamanlı KDK ile sağlanacak olması tez çalışmasının en önemli zorluğu, araştırma konusu ve bir diğer özgün değeridir. Ayrıca oldukça verimli motorlar olan FDAM'ların KDK ile üstün başarımlı olarak kontrol edilebilmesi ile yüksek performans ve minimum enerji tüketimi hedeflenmiş ve robotik alanında önemli bir araştırma konusu olan güç yönetimi hakkında önemli çıkarımlar sağlanmıştır.

Tevfik Yiğit

Fırat University · Institute of Graduate Studies in Science

2023

Hareketli robotlarMakine öğrenmesi yöntemleriİnsani robot

İnsansı robotların yürüme becerilerinin derin pekiştirmeli öğrenme algoritmalarıyla geliştirilmesi

İnsansı robotlar için sağlam bir hareket geliştirmek, onlarca yıldır araştırılan zorlu bir problemdir. Çeşitli yürüme yaklaşımları önerilmiş ve yürüme performansı önemli ölçüde geliştirilmiş olsa da, kararlılık konusunda hala beklentilerin gerisinde kalmaktadır. Pekiştirmeli öğrenme yaklaşımları için düşük yakınsama ve eğitim verimliliği, uygulamaları sınırlandırmaktadır. Bu sınırlamaların üstesinden gelmek için bu tez çalışmasında, Robotis-OP2 insansı robotuna dayalı olarak geleneksel yörünge üreteci kontrolör ve Derin Pekiştirmeli Öğrenme (DPÖ) ile birleştirilmiş etkili bir çerçeve önerilmiştir. Bu çerçeve, oluşturulan yürüyüş yörüngesi parametrelerinin optimizasyonu ve duruş dengeleme sisteminden oluşmaktadır. Webots simülatöründe DPÖ algoritmalarından Düello Çift Derin Q Ağı (Düello ÇDQA) kullanılarak yürüyüş parametreleri optimize edilmiştir. Duruş dengeleme sistemi için kalça stratejisi benimsenmiştir. Önerilen çerçeve ve Robotis-OP2 insansı robotunun kendi yürüme algoritması ile hem simülasyon hem de gerçek ortamda deneysel çalışmalar gerçekleştirilmiştir. Deneysel sonuçlar, robotun önerilen çerçeve ile robotun kendi yürüme algoritmasına göre düz yürüme görevinin daha kararlı bir şekilde gerçekleştirildiğini göstermiştir. Tez çalışması kapsamında daha sonra, robotun eğimli yüzeylerde kararlı bir şekilde yürüyebilmesi için PID kontrolör ve DPÖ kontrolörden oluşan iki ayrı yürüyüş dengeleme çerçevesi önerilmiştir. DPÖ kontrolör olarak DDPG (Derin Deterministik Politika Gradyanı) algoritması tercih edilmiştir. PID kontrolör ile gerçekleştirilen deneysel çalışmalarda robotun duruşu gerçek zamanlı olarak ayarlanarak gövde yunuslama açısının istenilen referans değerde olması sağlanmıştır. DDPG kontrolör ile robotun eğimli yüzeylerde dengeli yürüyüşünün sağlanabilmesi için robotun gövde yunuslama açısının sıralı hareket dizisi öğrenilmiştir. Deneysel sonuçlar, DPÖ kontrolörün PID kontrolöre göre daha kullanışlı olduğunu ve daha kararlı bir yürüme sağladığını göstermiştir.

Çağrı Kaymak

Fırat University · Fen Bilimleri Enstitüsü

2023

Master'sOpen AccessTR

Dört bacaklı mobil robot tasarımı ve kesir dereceli PID ile denetimi

Mobil robotlar, herhangi bir yere sabit olmadan üzerlerindeki sensörler vasıtasıyla karşısına çıkan engelleri algılayan, geliştirilen algoritma ve kontrol yöntemleriyle kendi başlarına öğrenme yeteneklerine sahip olmasından dolayı insan hayatını kolaylaştıran ve birçok alanda tercih edilen bir robot türüdür. Mobil robot kontrolünde başarılı sonuçlar elde etmek için matematiksel modelinin elde edilmesi ve uygun kontrol yönteminin seçilmesi gerekir. Bu tez çalışmasında, dört ayaklı iki serbestlik derecesine sahip mobil robotun kinematik ve dinamik denklemleri elde edilmiştir. Dört ayaklı mobil robotun, kinematik ve dinamik denklemleri elde edildikten sonra dört ayaklı mobil robotun kontrolünde kullanılacak olan kesir dereceli kontrol yöntemi ele alınmıştır. Ele alınan kesir dereceli kontrol yöntemi ile dört ayaklı mobil robotun yörünge kontrolünde kullanılacak olan DC motorun hız kontrolü gerçekleştirilmiştir. DC motorun, hız kontrolü gerçekleştirildikten sonra dört ayaklı mobil robotun katı model tasarımı Solidworks programından faydalanılarak gerçekleştirilmiştir. Solidworks programından katı model tasarımı gerçekleştirilen dört ayaklı mobil robotun, yörünge kontrolünü gerçekleştirmek için katı model tasarımı MATLAB ortamına aktarılmıştır. Elde edilen kinematik denklemler kullanılarak dört ayaklı mobil robotun yörünge kontrolü, MATLAB programından yararlanılarak kesir dereceli PID kontrol yöntemi ile gerçekleştirilmiştir. Ayrıca klasik PID kontrol yöntemi ile kesir dereceli PID kontrol yöntemi karşılaştırılmıştır ve kesir dereceli PID kontrol yöntemiyle tasarlanan sistemin belirlenen yörüngeyi daha az hatayla takip ettiği ve daha başarılı sonuçlar verdiği gözlemlenmiştir. Böylelikle Kesir dereceli PID kontrol yöntemiyle tasarlanan sistemin daha verimli çalıştığı gösterilmiştir.

Bacaklı robotlarHareketli robotlar

Abdulkadir Hanbay

Fırat University · Institute of Graduate Studies in Science

2024

Obstacle avoidanceMotion plannerMobile robots+1

Design, construction and control of an autonomous mobile rescue robot with visual feedback

The construction of a smart autonomous mobile robot system that can operate in an anonymous environment can be difficult [4]. Building an autonomous mobile robot requires collecting data from sensors and cameras in order to recognize the surrounding environment, get knowledge of existing landmarks, and figure out its position and orientation. The most common platform utilized for robot motion design is ROS (Robot Operating System) [29][35] and for object detection YOLO v3 (You Only Look Once, version 3) is a common choice [1]. Additionally, the autonomous system needs motion planning in a systematic way to avoid obstacles by using RPLIDAR A1 or other sensing methods generating the feedback signals required for control of motion around occupied spaces [2]. One of the most effective and versatile feedback systems for mobile robot control can be constructed using visual feedback, which relies on a high-resolution camera that processes the image in real-time and the result is fed back to the control system for decision making. In this research, results of design and construction of an autonomous mobile rescue robot for finding victims in hazardous disaster areas through video data are presented. It is aimed to find victims via applying video as a data collection and feedback generation tool in the autonomous mobile robot control system. The visual feedback and YOLO v3 are used not only for finding victims but also for motion control. Our autonomous mobile robot is designed to move smoothly to the victim's direction and avoids obstacles by using ROS Global Planner.

İbrahim Hasan

Gaziantep University · Institute of Graduate Studies in Science