Eliot Winer

Path Planning of Unmanned Aerial Vehicles using B-Splines and Particle Swarm Optimization

Military operations are turning to more complex and advanced automation technologies forminimumriskandmaximumefficiency.Acriticalpiecetothisstrategyisunmannedaerial vehicles. Unmanned aerial vehicles require the intelligence to safely maneuver along a path to an intended target and avoiding obstacles such as other aircrafts or enemy threats. This paper presents a unique three-dimensional path planning problem formulation and solution approach using particle swarm optimization. The problem formulation was designed with three objectives: 1) minimize risk owing to enemy threats, 2) minimize fuel consumption incurred by deviating from the original path, and 3) fly over defined reconnaissance targets. The initial design point is defined as the original path of the unmanned aerial vehicles. Using particle swarm optimization, alternate paths are generated using B-spline curves, optimized based on the three defined objectives. The resulting paths can be optimized with a preference toward maximum safety, minimum fuel consumption, or target reconnaissance. This method has been implemented in a virtual environment where the generated alternate paths can be visualized interactively to better facilitate the decision-making process. The problem formulation and solution implementation is described along with the results from several simulated scenarios demonstrating the effectiveness of the method.

THREE-DIMENSIONAL PATH PLANNING OF UNMANNED AERIAL VEHICLES USING PARTICLE SWARM OPTIMIZATION

Military operations are turning to more complex and advanced automation technology for minimum risk and maximum efficiency. A critical piece to this strategy is unmanned aerial vehicles (UAVs). UAVs require the intelligence to safely maneuver along a path to an intended target, avoiding obstacles such as other aircrafts or enemy threats. Often automated path planning algorithms are employed to specify targets for a UAV to fly to. To date, path-planning algorithms have been limited to two-dimensional problem formulations. This paper presents a unique three-dimensional path planning problem formulation and solution approach using Particle Swarm Optimization (PSO). The problem formulation was designed to minimize risk due to enemy threats while simultaneously minimizing fuel consumption. The initial design point is a straight path between the current position and the desired target. Using PSO, an optimized path is generated through B-spline curves. The resulting paths can be optimized with a preference towards maximum safety, minimum fuel consumption or a combination of the two. The problem formulation and solution implementation is described along with the results from several simulated scenarios.

Synchronous parallelization of Particle Swarm Optimization with digital pheromones

In this paper, Particle Swarm Optimization (PSO) using digital pheromones to coordinate swarms within n-dimensional design spaces in a parallel computing environment is presented. Digital pheromones are models simulating real pheromones emitted by insects for communication to indicate suitable food or nesting location. Particle swarms search the design space with digital pheromones aiding communication within the swarm during an iteration to improve search efficiency. Previous work by the authors demonstrated the capability of digital pheromones within PSO for searching the global optimum with improved accuracy, efficiency and reliability in a single processor computing environment. When multiple swarms explore and exploit the design space in a parallel computing environment, the solution characteristics can be further improved. This premise is investigated through deploying swarms on multiple processors in a distributed memory parallel computing environment. The primary hurdle for the developed algorithm was bandwidth latency due to synchronization across processors, causing the solution duration due to each swarm to be only as fast as the slowest participating processor. However, it has been observed that the speedup and parallel efficiency improved substantially as the dimensionality of the problems increased. The development of the method along with results from six test problems is presented.

A Multidimensional Visualization Interface to Aid in Trade-off Decisions During the Solution of Coupled Subsystems Under Uncertainty

In this paper, the application of visualization to aid in trade-off decisions when solving multi-objective optimisation problems involving coupled subsystems under uncertainty is presented. The developed visualization method is an abstract way of representing these problems for use in a solution process where multiple resultant design points may exist. In these cases, a designer must consider potentially complex trade-off decisions in order to choose a point with which to proceed. Designers can examine the coupled subsystem design space under different uncertainty conditions through three-dimensional visual representations. Promising solution regions can be determined and explored down to a specific design point. As a result, the time to locate a trade-off solution for all subsystems has the potential to substantially decrease. This paper presents background into the type of problems being addressed as well as other visualization methods used in design. The method development is presented along with the results and discussion of a test case to illustrate the methods viability.

Particle Swarm Optimization-Based Source Seeking

The task of locating a source based on the measurements of the signal emitted/emanating from it is called the source-seeking problem. In the past few years, there has been a lot of interest in deploying autonomous platforms for source-seeking. Some of the challenging issues with implementing autonomous source-seeking are the lack of a priori knowledge about the distribution of the emitted signal and presence of noise in both the environment and on-board sensor measurements. This paper proposes a planner for a swarm of robots engaged in seeking an electromagnetic source. The navigation strategy for the planner is based on Particle Swarm Optimization (PSO) which is a population-based stochastic optimization technique. An equivalence is established between particles generated in the traditional PSO technique, and the mobile agents in the swarm. Since the positions of the robots are updated using the PSO algorithm, modifications are required to implement the PSO algorithm on real robots to incorporate collision avoidance strategies. The modifications necessary to implement PSO on mobile robots, and strategies to adapt to real environments are presented in this paper. Our results are also validated on an experimental testbed. Note to Practitioners-This paper is inspired by the source seeking problem in which the signal emitted from the source is assumed to be very noisy, and the spatial distribution is assumed to be non-smooth. We focus our work specifically on electromagnetic sources. However, the strategies proposed in this paper are also applicable to other kinds of sources, for example, nuclear, radiological, chemical or biological. We develop a planner for a swarm of mobile agents that try to locate an unknown electromagnetic source. The mobile agents know their own positions and can measure the signal strength at their current location. They can share information among themselves, and plan for the next step. We propose a complete solution to ensure the effectiveness of PSO in complex environments where collisions may occur. We incorporate static and dynamic obstacle avoidance strategies in PSO to make it fully applicable to real-world scenario. We validate the proposed technique on an experimental testbed. As a part of our future work, we will extend the technique to locate multiple sources of different kinds.

Evaluating Mental Workload of Two-Dimensional and Three-Dimensional Visualization for Anatomical Structure Localization

Visualization of medical data in three-dimensional (3D) or two-dimensional (2D) views is a complex area of research. In many fields 3D views are used to understand the shape of an object, and 2D views are used to understand spatial relationships. It is unclear how 2D/3D views play a role in the medical field. Using 3D views can potentially decrease the learning curve experienced with traditional 2D views by providing a whole representation of the patients anatomy. However, there are challenges with 3D views compared with 2D. This current study expands on a previous study to evaluate the mental workload associated with both 2D and 3D views. Twenty-five first-year medical students were asked to localize three anatomical structures--gallbladder, celiac trunk, and superior mesenteric artery--in either 2D or 3D environments. Accuracy and time were taken as the objective measures for mental workload. The NASA Task Load Index (NASA-TLX) was used as a subjective measure for mental workload. Results showed that participants viewing in 3D had higher localization accuracy and a lower subjective measure of mental workload, specifically, the mental demand component of the NASA-TLX. Results from this study may prove useful for designing curricula in anatomy education and improving training procedures for surgeons.

Digital Pheromone Implementation of PSO with Velocity Vector Accelerated by Commodity Graphics Hardware

In this paper, a model for Graphics Processing Unit (GPU) implementation of Particle Swarm Optimization (PSO) using digital pheromones to coordinate swarms within ndimensional design spaces is presented. Particularly, the velocity vector computations are carried out on graphics hardware. Previous work by the authors demonstrated the capability of digital pheromones within PSO for searching n-dimensional design spaces with improved accuracy, efficiency and reliability in serial, parallel and GPU computing environments. The GPU implementation was limited to computing the objective function values alone. Modern GPUs have proven to outperform the number of floating point operations when compared to CPUs through inherent data parallel architecture and higher bandwidth capabilities. This paper presents a method to implement velocity vector computations on a GPU along with objective function evaluations. Three different modes of implementation are studied and presented - First, CPU-CPU where objective function and velocity vector are calculated on CPU alone. Second, GPU-CPU where objective function is computed on the GPU and velocity vector is computed on GPU. Third, GPU-GPU where objective function and velocity vector are both evaluated on the GPU. The results from these three implementations are presented followed by conclusions and recommendations on the best approach for utilizing the full potential of GPUs for PSO.

Implementation of Digital Pheromones in Particle Swarm Optimization for Constrained Optimization Problems

This paper presents a model for digital pheromone i mplementation of Particle Swarm Optimization (PSO) to solve constrained optimization problems. Digital pheromones are models simulating real pheromones produced by insec ts for communication to indicate a source of food or a nesting location. When integrat ed within PSO, this principle of communication and organization between swarm member s offer substantial improvement in search accuracy, efficiency and reliability. Multiple pheromones are released in the design space, and the strength of a pheromone in a region of the design space is determined through empirical proximity analysis, and. The swarm then reacts accordingly based on the probability that this region may contain an optimum. The addition of a pheromone component to the velocity vector equation demonstra ted substantial success in solving unconstrained problems. The research presented in t his paper explores the suitability of the developed method to solve constrained optimization problems. A sequential unconstrained minimization technique – Augmented Lagrange Multiplier (ALM) method has been implemented to address constrained optimization pro blems. ALM has been chosen because of its relative insensitivity to whether the initial design points for a pseudo objective function are feasible or infeasible. The development of the method and results from solving several constrained test problems are presented.

A multi-fidelity software framework for interactive modeling of advective and diffusive contaminant transport in groundwater

Groundwater currently accounts for over 20% of the daily water usage in the United States. As the contamination of groundwater reserves also continues to increase from a variety of pollutants, remediation becomes necessary. Whether the remediation method involves containing or cleaning, a fundamental understanding of the groundwater flow patterns is necessary. This can now be gained through predictive models due to advances in groundwater simulation research. One such method, the Superblock Analytical Element Method (AEM) is capable of capturing both large-scale trends and small-scale variations in complex, heterogeneous flow fields. These trends and variations are not captured by current numerical solutions, typified by finite difference and finite element formulations. The Superblock AEM, as with many groundwater solvers, has a substantial number of input parameters and produces large amounts of 3D output. Thus, an intuitive, 3D visual framework would greatly enhance the usability of the method. With the Superblock AEM as a core solver, Groundwater TRANsport 3D (GTRAN3D) has been developed to provide a platform for modeling and viewing advective and diffusive contaminant spreading over non-intersecting spheroidal in-homogeneities in groundwater. The developed system allows a user to create an input scenario through an interactive three-dimensional graphical user interface (GUI), send that input to the solver, and then view the results on systems ranging from desktop computers to immersive Virtual Reality (VR) environments. In addition, the software can be used on any operating system or even accessed via a web interface. The development of the software framework is discussed along with the presentation of several test cases.

A Framework for Interactive Visualization of Digital Medical Images

The visualization of medical images obtained from scanning techniques such as computed tomography and magnetic resonance imaging is a well-researched field. However, advanced tools and methods to manipulate these data for surgical planning and other tasks have not seen widespread use among medical professionals. Radiologists have begun using more advanced visualization packages on desktop computer systems, but most physicians continue to work with basic two-dimensional grayscale images or not work directly with the data at all. In addition, new display technologies that are in use in other fields have yet to be fully applied in medicine. It is our estimation that usability is the key aspect in keeping this new technology from being more widely used by the medical community at large. Therefore, we have a software and hardware framework that not only make use of advanced visualization techniques, but also feature powerful, yet simple-to-use, interfaces. A virtual reality system was created to display volume-rendered medical models in three dimensions. It was designed to run in many configurations, from a large cluster of machines powering a multiwalled display down to a single desktop computer. An augmented reality system was also created for, literally, hands-on interaction when viewing models of medical data. Last, a desktop application was designed to provide a simple visualization tool, which can be run on nearly any computer at a users disposal. This research is directed toward improving the capabilities of medical professionals in the tasks of preoperative planning, surgical training, diagnostic assistance, and patient education.