Ph.D. Theses
Permanent URI for this collection
Browse
Browsing Ph.D. Theses by Author "Akın, H. Levent."
Now showing 1 - 6 of 6
Results Per Page
Sort Options
Item A collaborative multi-robot localization technique for autonomous robots(Thesis (Ph.D.)-Bogazici University. Institute for Graduate Studies in Science and Engineering, 2007., 2007.) Bağcı, Hatice Köse.; Akın, H. Levent.This work proposes a novel method for collaborative global localization of a team of soccer playing autonomous robots. It is also applicable to other indoor real-time robot applications in noisy, unpredictable environments, with insufficient perception. A novel solution, Reverse Monte Carlo Localization (R-MCL) is designed to solve single self-localization problem using local perception and action about the surrounding environment for each robot. R-MCL is a hybrid method based on Markov Localization (ML) and Monte Carlo Localization (MCL) where the ML based part finds the region where the robot should be and the MCL based part predicts the geometrical location with high precision by selecting samples in this region. In the multi-robot localization problem, robots use their own local position estimations, and the shared information from other team mates, to localize themselves. To integrate the local information and beliefs optimally, avoid conflicts and support collaboration among team members, a novel collaborative multi-robot localization method called Collaborative Reverse Monte Carlo Localization (CR-MCL), based on R-MCL, is presented. When robots detect each other, they share the grid cells representing this observation. The power of the method comes from its hybrid nature. It uses a grid based approach to handle detections which can not be accurate in real-time applications, and sample based approach in self-localization to improve its success, although it uses lower amount of samples compared to similar methods. Both methods are tested using simulated robots and real robots and results show that they are fast, robust, accurate and cheap in terms of communication, memory and computational costs.Item ADES : automatic driver evaluation system(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2011., 2011 .) Kaplan, Kemal.; Akın, H. Levent.Most of the traffic accidents occurred in the world and in our country are caused by the drivers. ADES (Automatic Driver Evaluation System) project targets to present a framework for integrating different applications for driver evaluation purpose. The proposed system can be divided into two main modules. The first one, which is the data acquisition and processing module, acquires the sensor information from the outside world and processed this data to present valuable information to the decision system. The system may benefit from built-in sensors like cameras or GPS (Global Positioning System) systems as well as non standard devices like RFID (Radio Frequency Identification) readers. The second module is the inference engine, which processes the information provided by the first module and makes judgments about the actions of the driver. Two sample expert system designs are proposed in the project. The developed solution is tested in simulation environment and by using real video recordings.Item Case-based mobile manipulation(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2014., 2014.) Meriçli, Tekin.; Akın, H. Levent.; Veloso, Manuela.The ability to manipulate the environment is one of the primary skills that autonomous mobile service robots are expected to have, considering that the daily lives of humans heavily rely on this skill. There are various ways for a mobile robot to perform manipulation, the exact form of which is determined by the requirements of the task and the constraints imposed by the physical properties of the environment, the object, and the robot itself. Anecdotal evidence suggests that humans mostly reuse their manipulation experiences, acquired through interaction and observation, especially in recurring everyday manipulation tasks, both in prehensile and non-prehensile manipulation contexts. With this motivation, this thesis contributes a case-based approach to achieving practical and e cient mobile manipulation through the utilization of past experience, stored as object-speci c, distinct, and potentially probabilistic cases. In scenarios where prehensile manipulation is possible, this guidance combined with sampling-based generative planners helps reduce planning time by deliberately biasing the planning process towards the feasible cases while increasing the overall robustness and repeatability of the method. When non-prehensile manipulation techniques, such as push-manipulation, need to be utilized, these probabilistic cases can be used as building blocks for constructing safe and achievable push plans to navigate the object of interest to the desired goal pose as well as to potentially push the movable obstacles out of the way in cluttered task environments. Additionally, incremental acquisition and tuning of the probabilistic cases allows the robot to adapt to the changes in the environment, such as increased mass due to loading of the object of interest for transportation purposes. The purely interaction and observation driven nature of our method makes it robot, object, and environment (real or simulated) independent, as we demonstrate through extensive testing and experimentation. We also verify the validity of our push-manipulation method in preliminary real world tests.Item Digital map and GNSS fusion to enhance localization for intelligent vehicle applications(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2017., 2017.) Peker, Ali Ufuk.; Akın, H. Levent.; Tosun, Oğuz.This thesis aims to enhance vehicle localization through the fusion of digital maps and vehicular communication. A particle lter based algorithm for fusing global navigation satellite system (GNSS) receiver, odometer, and digital maps is proposed and implemented. Implementation deployed on an embedded system and tested in the eld. Field tests were carried out di erent in parts of _Istanbul to measure the performance of the algorithm in di erent satellite visibility conditions. Results show that algorithm selects the correct road segment on the digital map with 96% success rate. The proposed algorithm was further enhanced with the addition of mutual positioning on vehicular ad-hoc networks (VANETs). The measurement-based statistical model of relative distance as a function of Time-of-Arrival(TOA) is experimentally obtained. The mutual positioning procedure is investigated in terms of positioning accuracy and network performance through realistic simulation studies with a di erent number of collaborative vehicles, and the proposed mutual positioning procedure is experimentally evaluated by a eet of ve IEEE 802.11p radio modem equipped vehicles. It is shown that collaboration in a VANET improves the availability of position measurement and its accuracy up to 40% in comparison with the stand-alone GNSS receiver. Local integrity heat map concept is introduced as a new local integrity methodology. Local integrity heat map is implemented and tested with extensive eld tests. Algorithm is successful in detecting urban canyons and can be used as an augmentation for GNSS. It is concluded that our fusion framework with the use of digital maps and inter-vehicle communications can be e ciently used for ADAS applications and our local integrity method can further enhance fusion and localization.Item Evolutionary algorithms for solving DEC-POMDP problems(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2012., 2012.) Eker, Barış.; Akın, H. Levent.The Decentralized Partially Observable Markov Decision Process (DEC-POMDP) model addresses the multiagent planning problem in partially observable environments. Due to its NEXP-complete complexity, only small size problems can be solved exactly. For this reason, many researchers concentrate on approximate solution algorithms that can handle more complex cases and produce near optimal solutions. However, even the approximate solution techniques developed so far can handle large size problems only for small horizons. One reason for this is the exponential memory requirements while representing the agent policies and searching the policy space. In this thesis, we propose four new approaches to solve finite horizon DEC-POMDP problems approximately. The first approach, called MAP, is based on modeling DEC-POMDP problems as a POMDP problem and then solving using an efficient POMDP solver. The other approaches, namely ES-BV, ES-OH and GA-FSC, are all based on the application of evolutionary algorithms. The ESBV makes use of belief vectors as in the case of MAP and tries to find policy vectors using evolution strategies (ES). The ES-OH proposes to use the observation history and input it into a neural network to make a decision and it uses ES to train the neural networks. The GA-FSC algorithm makes use of finite state controllers for representing the policies and search for the optimal policy using genetic algorithms (GA). All algorithms were tested on the major well-known DEC-POMDP problems. We compared our results with the current state of the art methods and we also compared our algorithms with each other. We showed that all the algorithms developed in this study, except MAP, have comparable performance to that of the existing top algorithms and in the case of the GA-FSC, the solution horizon for the problems are extended at least an order of magnitude.Item Multi-resolution model plus correction paradigm for task and skill refinement on autonomous robots(Thesis (Ph.D.) - Bogazici University. Institute for Graduate Studies in Science and Engineering, 2011., 2011.) Meriçli, Çetin.; Akın, H. Levent.; Veloso, Manuela.Robots need to be taught what type of tasks or skills they are expected to perform, and how to perform those particular tasks or skills. However, there is no universally accepted single approach for transferring the task and skill knowledge to a robot. Among several popular approaches, the most widely adopted method for transferring the task or skill knowledge to the robot is to develop an algorithm for performing the task or skill in question. Such a development requires a model of the system to be available. Moreover, despite that it usually is easier to develop a simple algorithm to handle trivial cases, it becomes a time consuming process to keep re ning the algorithm by modifying the underlying model to handle more complex situations. Learning from Demonstration (LfD) is another popular approach for transferring the task and skill knowledge to the robot. Instead of explicit programming, a teacher demonstrates the robot how to perform the task or skill and the robot records the demonstrated action together with the perceived state of the system at the time of demonstration. An execution policy is then derived out of the recorded demonstration data for reproducing the task or skill. Depending on the complexity of the task or skill in question and the robotic platform to be used, providing su cient number of examples in order to be able to extract a generalized execution policy can be a very time consuming process. This thesis contributes a novel complementary corrective demonstration paradigm called Model Plus Correction (M+C) for task and skill re nement on autonomous robots. The M+C approach strikes a balance between model-based and data-driven methods by combining them in a complementary manner. We assume the availability of an algorithm capable of performing the task or skill in question with limited success in terms of performance. Our approach utilizes a human teacher who observes the partially successful execution of the task, and corrects the action of the robot when the default algorithm is unable to select an appropriate action to be executed. The collected demonstration data stamped with the state of the system at the time of demonstration is then used to augment the default algorithm by modifying the action computed by the algorithm according to a correction reuse function, and the state of the system. This thesis also introduces an algorithm for using the same complementary corrective demonstration approach at multiple detail resolutions. The Multi-Resolution Model Plus Correction (MRM+C) algorithm assumes that a set of detail levels are de- ned with di erent state and action representations together with a di erent modelbased controller for each detail level are available at hand. The teacher provides demonstration for which detail resolution to use at a particular state of the system in addition to delivering corrective demonstration for the controller associated with the current detail resolution. Having multiple detail resolutions with di erent complexities allows the system to use more detailed state and action representations and more complex model-based controllers only when needed. Using a less detailed state and action representation with a simpler controller makes it possible to cover the solution space at a lower computational cost and using fewer number of demonstrations. The learned detail resolution selection policy favors the least detailed resolution by default and switches to a more detailed resolution if commanded to do so in a similar state before. We present experiment results where the M+C approach is rst applied to a complex biped walk stability improvement problem as an example to the skill refinement, and to a ball dribbling problem in a robot soccer environment as an example to the task re nement. We also present experiment results where the MRM+C approach is applied to a humanoid obstacle avoidance task on a robot soccer eld. Finally, we present an experimental analysis of the proposed algorithms in terms of their robustness against uncertainty and the cost analysis of using multiple detail resolutions over using a single detail resolution in a simulated version of the obstacle avoidance task.