Low-infrastructure techniques for indoor localization attempt to provide indoor positioning information for users, without requiring the installation of specialized transmitting or receiving hardware. Such an approach should encourage further adoption of indoor positioning systems by reducing the installation burden on individual building owners. If fully adopted, indoor positioning could prove to be a valuable addition to the existing outdoor localization system based on GPS. In this work, a particle filter is used to combine motion and light data in order to provide positioning information for a user in an indoor environment. A simple lighting model is used to predict light measurements, while an orientation tracking algorithm provides information about user motion. The system is shown to work with the existing lighting infrastructure of a building, though the addition of visible light communication (VLC) enabled light fixtures is shown to further improve performance. An experimental demonstration of the proposed system is provided, which indicates that tracking accuracy on the order of ten’s of centimetres is possible with very low infrastructure requirements. / Thesis / Master of Applied Science (MASc)

No description available.

Localization is an essential part of any autonomous vehicle. In a simple setting, the localization problem is almost trivial, and can be solved sufficiently using simple dead reckoning or an off-the-shelf GPS with differential corrections. However, as the surroundings become more complex, so does the localization problem. The urban environment is a prime example of a situation in which a vehicle's surroundings complicate the problem of position estimation. The urban setting is marked by tall structures, overpasses, and tunnels. Each of these can corrupt GPS satellite signals, or completely obscure them, making it impossible to rely on GPS alone. Dead reckoning is still a useful tool in this environment, but as is always the case, measurement and modeling errors inherent in dead reckoning systems will cause the position solution to drift as the vehicle travels eventually leading to a solution that is completely diverged from the true position of the vehicle. The most widely implemented method of combining the absolute and relative position measurements provided by GPS and dead reckoning sensors is the Extended Kalman Filter (EKF). The implementation discussed in this paper uses two Kalman Filters to track two completely separate position solutions. It uses GPS/INS and odometry to track the Absolute Position of the vehicle in the Global frame, and simultaneously uses odometry alone to compute the vehicle's position in an arbitrary Local frame. The vehicle is then able to use the Absolute position estimate to navigate on the global scale, i.e. navigate toward globally referenced checkpoints, and use the Relative position estimate to make local navigation decisions, i.e. navigating around obstacles and following lanes. This localization solution was used on team VictorTango's 2007 DARPA Urban Challenge entry, Odin. Odin successfully completed the Urban Challenge and placed third overall. / Master of Science

In this thesis, we build a robust end-to-end pipeline for the localization of multiple drones in a GPS-denied environment. This pipeline would help us with cooperative formation control, autonomous delivery, search and rescue operations etc. To achieve this we integrate a custom trained YOLO (You Only Look Once) object detection network, for drones, with the ZED2 stereo camera system. With the help of this sensor we obtain a relative vector from the left camera to that drone. After calibrating it from the left camera to that drone's center of mass, we then estimate the location of all the drones in the leader drone's frame of reference. We do this by solving the localization problem with least squares estimation and thus acquire the location of the follower drone's in the leader drone's frame of reference. We present the results with the stereo camera system followed by simulations run in AirSim to verify the precision of our pipeline. / Master of Science / In the recent years, technologies like Deep Learning and Machine Learning have seen many rapid developments. This has lead to the rise of fields such as autonomous drones and their application in fields such as bridge inspection, search and rescue operations, disaster management relief, agriculture, real estate etc. Since GPS is a highly unreliable sensor, we need an alternate method to be able to localize the drones in various environments in real time. In this thesis, we integrate a robust drone detection neural network with a camera which estimates the location. We then use this data to get the relative location of all the follower drones from the leader drone. We run experiments with the camera and in a simulator to show the accuracy of our results.

Multi-agent robotic system have been proved to be more superior in undertaking functionalities, arduous or even impossible when performed by single agents. The increased efficiency in multi agent systems is achieved by the execution of the task in cooperative manner. But to achieve cooperation in multi agent systems, a good localization system is an important prerequisite. Currently, most of the multi-agent system rely on the use of the GPS to provide global positioning information which suffers great deterioration in performance in indoor applications, and also all to all communication between the agents will be required which is not efficient especially when the number of agents is large. In this regard, a real-time localization scheme is introduced which makes use of the robot’s on-board sensors and computational capabilities to determine the states of other agents in the multi agent system. This algorithm also takes the advantage of the swarming behaviour of the robots in the estimation of the states. This localization algorithm was found to produce more accurate agent state estimates as compared to a similar localization algorithm that does not take into account the swarming behaviour of the agents in simulations and real experiment involving two Unmanned Aerial Vehicles.

The single hydrophone localization problem is considered. Single hydrophone localization is a special case of matched field localization where measurements from only one hydrophone are available. The time series of the pressure at the hydrophone is compared with predicted times series calculated using an ocean acoustic propagation model for many different source locations. The source location that gives the best match between the predicted time series and the measurement is assumed to be the correct source location. Single hydrophone localization algorithms from the literature are reviewed and a new algorithm is introduced. The new algorithm does not require knowledge of the source signal and does not assume the use of a particular ocean acoustic model, unlike some algorithms in the literature. Source location estimates calculated from the new algorithm are compared with ground truth using simulated ocean acoustic measurements and experimental measurements. Source location estimates calculated using other algorithms from the literature are shown for comparison. The simulated measurements use three source signals with bandwidths of 10 Hz, 100 Hz, and 200 Hz and the ocean is modeled as a Pekeris waveguide. The new algorithm estimates the source location accurately for all three source signals when several of the localization algorithms from the literature give inaccurate estimates. Gaussian white noise signals are added to the measured signals to test the impact of signal-to-noise ratio (SNR) on the algorithm. Four signal-to-noise ratios of 60 dB, 40 dB, 20 dB, and 0 dB are used. The new algorithm gives accurate source location estimates down to an SNR of 20 dB for two of the source signal bandwidths. Source location estimates using other algorithms from the literature break down at either 20 dB or 0 dB. Source location estimates are calculated using two hydrophone measurements taken at different depths in an experiment conducted near the Bahamas. The new algorithm accurately estimates the source location in both cases. In one case, only two other localization algorithms from the literature locate the source accurately. In the other case, only one other localization algorithm succeeds.

This thesis is focused on definition of basic factors of localization of brewing industry in Czech Republic. Traditionally, beer is the part of Czech culture. The aim of this thesis is to identify main factors of localization, to analyze them and to evaluate their influence on spatial distribution of brewing industry in Czech Republic. Thesis is divided into four parts. The first part is focused on summary of basic theoretical knowledge of localization. The second part summarizes important facts about beer including history, production process and basic statistics of food and beverage and brewing industry. The third part analyzes spatial distribution of brewing industry in Czech Republic. The fourth part is focused on analysis and evaluation of the importance of factors of localization. In the end of thesis are summarized the results.

Wireless sensor networks allow us to instrument our world in novel ways providing detailed insight that had not been possible earlier. Since these networks provide an interface to the physical world, it is necessary for each sensor node to learn its location in the physical space. The availability of location information at individual sensor nodes allows the network to provide higher layer services such as location stamped event reporting, geographic routing, in-network processing etc. A wide range of these sensor network protocols do not require absolute node coordinates and can work with relative node positions. This motivates for the need of anchor free localization algorithms that localize the individual sensor nodes with respect to each other in a local coordinate system. Such algorithms allow the sensor networks to be decoupled from external infrastructure and become truly place and play systems. The primary contributions of this thesis include two anchor free localization algorithms and one location refinement algorithm for ad-hoc wireless sensor networks. Our distributed anchor free localization algorithms do not require any external infrastructure in the form of landmark or manually initialized anchor nodes. These algorithms use measured inter-node distances among some node pairs and localize the entire network in a local coordinate system up to a global translation, rotation and reflection. The relative or virtual coordinates assigned by these algorithms can be readily used with a range of sensor network services like geographic routing, data aggregation, topology control etc. Our first localization algorithm is based on a distributed collaborative approach where all of the nodes in the network collaborate with each other to select a set of nodes. These nodes are localized and then used as reference nodes for the remaining sensor nodes. The novelty of this approach is that instead of solving the localization problem for the entire network upfront, first a small well-formed localization problem is solved and then these results are used to solve the localization problem for the remaining nodes in the network. Our second localization algorithm borrows ideas from the data visualization field and exploits the general undirected graph drawing theory to solve the sensor network localization problem. This algorithm divides the network into a large number of small overlapping clusters and creates local coordinate systems for each of the clusters. These clusters are then merged together in a single coordinate system using a novel distributed algorithm that seeks to minimize the error during this merge process. Our final contribution is a distributed location refinement algorithm that can be used with any of the range based localization algorithms to refine the sensor node coordinates to conform to the measured inter-node distances. We model this coordinate refinement problem as an unconstrained non-linear optimization problem and then transform this optimization problem into an aggregate computation problem. We propose two different approaches to solve this aggregate computation problem in a distributed manner. We evaluate our algorithms with detailed simulations using both Matlab and TinyOS simulator TOSSIM. We also validate our simulation results with experimentation carried out on a real network of MIT Cricket motes. We conclude this thesis with lessons learned during this research and discuss some future directions which can be explored to advance the research in sensor network localization.

In this thesis, we develop a method using fuzzy logic to do cooperative localization. In a group of robots, at a given instant, each robot gives crisp pose estimates for all the other robots. These crisp pose values are converted to fuzzy membership functions based on various physical factors like acceleration of the robot and distance of separation of the two robots. For a given robot, all these fuzzy estimates are taken and fused together using fuzzy fusion techniques to calculate a possibility distribution function of the pose values. Finally, these possibility distributions are defuzzified using fuzzy techniques to find a crisp pose value for each robot. A MATLAB code is written to simulate this fuzzy logic algorithm. A Kalman filter approach is also implemented and then the results are compared qualitatively and quantitatively.

In this thesis, we develop a method using fuzzy logic to do cooperative localization. In a group of robots, at a given instant, each robot gives crisp pose estimates for all the other robots. These crisp pose values are converted to fuzzy membership functions based on various physical factors like acceleration of the robot and distance of separation of the two robots. For a given robot, all these fuzzy estimates are taken and fused together using fuzzy fusion techniques to calculate a possibility distribution function of the pose values. Finally, these possibility distributions are defuzzified using fuzzy techniques to find a crisp pose value for each robot. A MATLAB code is written to simulate this fuzzy logic algorithm. A Kalman filter approach is also implemented and then the results are compared qualitatively and quantitatively.