Browsing by Subject "Robot"
Now showing 1 - 11 of 11
Results Per Page
Sort Options
Item Autonomous robotic wheelchair with collision-avoidance navigation(Texas A&M University, 2008-10-10) Hsieh, Pin-ChunThe objective of this research is to demonstrate a robotic wheelchair moving in an unknown environment with collision-avoidance navigation. A real-time path-planning algorithm was implemented by detecting the range to obstacles and by tracking specific light sources used as beacons. Infrared sensors were used for range sensing, and light-sensitive resistors were used to track the lights. To optimize the motion trajectory, it was necessary to modify the original motor controllers of the electrical wheelchair so that it could turn in a minimum turning radius of 28.75 cm around its middle point of axle. Then, with these kinematics, the real-time path planning algorithm of the robotic wheelchair is simplified. In combination with the newly developed wireless Internet-connection capability, the robotic wheelchair will be able to navigate in an unknown environment. The experimental results presented in this thesis include the performance of the control system, the motion trajectory of the two driving wheels turning in a minimum radius, and the motion trajectory of the real-time path-planning in a real-life testing environment. These experimental results verified that the robotic wheelchair could move successfully in an unknown environment with collision-avoidance navigation.Item A compliant control law for industrial, dual-arm manipulators(2013-05) Zelenak, Andrew J; Landsberger, Sheldon; Pryor, Mitchell WayneMany of the first robots ever built, decades even before the first industrial robots, were humanoids. It seems like researchers have always sought to imitate the human form with their robots, and with good reason. Humans are incredibly flexible; they can perform a huge variety of tasks, from locomotion over rough terrain, to delicate assembly, to heavy lifting. A human’s second arm allows him to lift twice as much weight. His workspace is approximately doubled, and he can perform a broader variety of tasks as items are passed back and forth between hands. We sought to impart some of that same functionality to a strong, rigid, dual-arm robot. Specifically, we developed a control law that allows two robot arms to lift and manipulate an object in cooperation. As opposed to the prior art, our control law is tailored for industrial robots. These robots do not usually allow torque control and their control frequency is generally 60 Hz. Through the use of fuzzy logic, the control law is quite robust at 60 Hz control rates. Its simple structure reduces the computational cost of the algorithm by approximately 75% over Jacobian-based methods. Stability is proven and the controller parameters can be adjusted to handle perturbances of arbitrary magnitude. Since the robots behave as an admittance, torque control is not required. Several experiments were conducted to benchmark and validate the performance of this control law. The controller is able to maintain a clamp force within ± 4N despite a wide variation in trajectory and control frequency. This fine level of force control makes the controller suitable for delicate tasks. The conclusion suggests several extensions that would make this control law more useful. For example, adaptive control would improve the performance. A position feedback controller should be cascaded so that the robot arms’ tracking accuracy is improved. Many tasks (such as co-robotics) require external compliance, and we show how external compliance could easily be incorporated.Item Feedback-based Information Roadmap (FIRM): Graph-based Estimation and Control of Robotic Systems Under Uncertainty(2014-05-07) Aghamohammadi, AliakbarThis dissertation addresses the problem of stochastic optimal control with imperfect measurements. The main application of interest is robot motion planning under uncertainty. In the presence of process uncertainty and imperfect measurements, the system's state is unknown and a state estimation module is required to provide the information-state (belief), which is the probability distribution function (pdf) over all possible states. Accordingly, successful robot operation in such a setting requires reasoning about the evolution of information-state and its quality in future time steps. In its most general form, this is modeled as a Partially-Observable Markov Decision Process (POMDP) problem. Unfortunately, however, the exact solution of this problem over continuous spaces in the presence of constraints is computationally intractable. Correspondingly, state-of-the-art methods that provide approximate solutions are limited to problems with short horizons and small domains. The main challenge for these problems is the exponential growth of the search tree in the information space, as well as the dependency of the entire search tree on the initial belief. Inspired by sampling-based (roadmap-based) methods, this dissertation proposes a method to construct a "graph" in information space, called Feedback-based Information RoadMap (FIRM). Each FIRM node is a probability distribution and each FIRM edge is a local controller. The concept of belief stabilizers is introduced as a way to steer the current belief toward FIRM nodes and induce belief reachability. The solution provided by the FIRM framework is a feedback law over the information space, which is obtained by switching among locally distributed feedback controllers. Exploiting such a graph in planning, the intractable POMDP problem over continuous spaces is reduced to a tractable MDP (Markov Decision Process) problem over the graph (FIRM) nodes. FIRM is the first graph generated in the information space that preserves the principle of optimality, i.e., the costs associated with different edges of FIRM are independent of each other. Unlike the forward search methods on tree-structures, the plans produced by FIRM are independent of the initial belief (i.e., plans are query-independent). As a result, they are robust and reliable. They are robust in the sense that if the system's belief deviates from the planned belief, then replanning is feasible in real-time, as the computed solution is a feedback over the entire belief graph. Computed plans are reliable in the sense that the probability of violating constraints (e.g., hitting obstacles) can be seamlessly incorporated into the planning law. Moreover, FIRM is a scalable framework, as the computational complexity of its construction is linear in the size of underlying graph as opposed to state-of-the-art methods whose complexity is exponential in the size of underlying graph. In addition to the abstract framework, we present concrete FIRM instantiations for three main classes of robotic systems: holonomic, nonholonomic, and non-pointstabilizable. The abstract framework opens new avenues for extending FIRM to a broader class of systems that are not considered in this dissertation. This includes systems with discrete dynamics or in general systems that are not well-linearizable, systems with non-Gaussian distributions, and systems with unobservable modes. In addition to the abstract framework and concrete instantiations of it, we propose a formal technique for replanning with FIRM based on a rollout-policy algorithm to handle changes in the environment as well as discrepancies between actual and computational models. We demonstrate the performance of the proposed motion planning method on different robotic systems, both in simulation and on physical systems. In the problems we consider, the system is subject to motion and sensing noise. Our results demonstrate a significant advance over existing approaches for motion planning in information space. We believe the proposed framework takes an important step toward making information space planners applicable to real world robotic applications.Item Implementation of multi-algorithm controllers for path determination in mobile robot systems(2006-05) Hitijahubessy, Adrianus Victor; Fernandez, Benito R.Recent advancements in control systems, such as the ones used in missile technology in the military or autonomous vehicle development have motivated this study in an attempt to explore various control algorithms and their implementation relevant those applications. Both missile interceptor and autonomous vehicle technology require precise and responsive control system to accurately determine the projectile path of pursuer to strike a moving target or reach a static finish line.The objective of this study is to investigate the performance of several control techniques for a mobile robot to autonomously track and pursue a moving object. Computer model is developed to numerically predict the path taken by the pursuer as it tracks an object moving in regular or random manner. In the computer simulation, the robot's path is calculated using three different techniques: reactive controller, linear estimation, and artificial neural network. Fitness of each method may be determined by evaluating the controller against several factors, such as interception time, steady-state positional error, steady-state time (settling time) and algorithm complexity, listed in decreasing order of importance. A working experimental model is developed to validate the controller selection determined from the computer model simulation. In the experimental setting, the primary inputs to the robot are visual images from cameras. The experiments are carried out with the robot receiving visual inputs from two different perspectives, overhead and frontal vision. Robust image processing technique becomes a topic of significant importance for the system. To manipulate visual images in real-time from raw inputs to comprehensible data, while maintaining fast computational time is a challenge that is addressed in this study. The results from computer simulations show that artificial neural network is a more powerful control algorithm, capable of estimating the object's path more accurately than the other two controllers, resulting in smaller steady-state positional error. The experimental results confirm this conclusion as artificial neural network outperforms the reactive and linear controller by intercepting the object more quickly, i.e. shorter interception time.Item Learning in simulation for real robots(2012-05) Farchy, Alon; Stone, Peter, 1971-; Ballard, Dana H.Simulation is often used in research and industry as a low cost, high efficiency alternative to real model testing. Simulation has also been used to develop and test powerful learning algorithms. However, optimized values in simulation do not translate directly to optimized values in application. In fact, heavy optimization in simulation is likely to exploit the simplifications made in simulation. This observation brings to question the utility of learning in simulation. The UT Austin Villa 3D Simulation Team developed an optimization framework on which a robot agent was trained to maximize the speed of an omni-directional walk. The resulting agent won first place in the 2011 RoboCup 3D Simulation League. This thesis presents the adaptation of this optimization framework to learn parameters in simulation that improved the forward walk speed of the real Aldebaran Nao. By constraining the simulation with tree models learned from the real robot, and manually guiding the optimization based on testing on the real robot, the Nao's walk speed was improved by about 30%.Item Monte Carlo localization for robots using dynamically expanding occupancy grids(Texas Tech University, 2005-05) Gupta, Karan M.; Pyeatt, Larry D.; Watson, RichardThe past few years have seen tremendous growth in the research areas of Mobile Robotics. While growth has been fast and several problems have been very splendidly solved most mobile roboticists are faced with two primary challenges: how will the robot gather information about its environment and how will it know where it is? These two problems are referred to as:
(i). Mapping and
(ii). Localization.
Mapping is the process whereby a robot can extract relevant information from its environment allowing it to ”remember” it. Localization is using this stored map to move about in the environment with a clear sense of direction because the robot knows where it is, by referring to the map. Localization is the problem of estimating a robot’s pose relative to a map of its environment. However, both these problems are computationally intensive to solve and furthermore, limitations on a robot’s on board computational abilities and inaccuracies in sensor hardware and motor effectors make it even harder. Most mapping techniques are limited by memory and hence a robot has a limitation on the area that it can directly map. Also, if the mapped area is extended, most mapping implementations require that the mapping parameters be changed and the entire mapping algorithm be executed again. However, in recent times a new mapping technique was explored which is that of using Dynamically Expanding Occupancy Grids (Ellore 2002), and of using a Centralized Storage System (Barnes, Quasny, Garcia, and Pyeatt 2004). By using this approach, the robot has virtually unlimited storage space and a small initial map which grows as the robots explores its environment. Localization has not yet been attempted using Dynamically Expanding Occupancy Grids and a Centralised Storage System. This research is geared towards implementing Monte-Carlo Localization methods (Fox, Burgard, Dellaert, and Thrun 1999; Dellaert, Fox, Burgard, and Thrun ; Thrun, Fox, Burgard, and Dellaert 2001; Fox, Thrun, Burgard, and Dellaert 2001) to robots using Dynamically Expanding Occupancy Grids. By using this approach this research aims to provide a complete mapping and localization implementation for robots using dynamically expanding occupancy grids and a centralized storage system.Item Monte Carlo localization for robots using dynamically expanding occupancy grids(2005-05) Gupta, Karan M.; Pyeatt, Larry D.; Watson, RichardThe past few years have seen tremendous growth in the research areas of Mobile Robotics. While growth has been fast and several problems have been very splendidly solved most mobile roboticists are faced with two primary challenges: how will the robot gather information about its environment and how will it know where it is? These two problems are referred to as: (i). Mapping and (ii). Localization. Mapping is the process whereby a robot can extract relevant information from its environment allowing it to "remember" it. Localization is using this stored map to move about in the environment with a clear sense of direction because the robot knows where it is, by referring to the map. Localization is the problem of estimating a robot’s pose relative to a map of its environment. However, both these problems are computationally intensive to solve and furthermore, limitations on a robot’s on board computational abilities and inaccuracies in sensor hardware and motor effectors make it even harder. Most mapping techniques are limited by memory and hence a robot has a limitation on the area that it can directly map. Also, if the mapped area is extended, most mapping implementations require that the mapping parameters be changed and the entire mapping algorithm be executed again. However, in recent times a new mapping technique was explored which is that of using Dynamically Expanding Occupancy Grids (Ellore 2002), and of using a Centralized Storage System (Barnes, Quasny, Garcia, and Pyeatt 2004). By using this approach, the robot has virtually unlimited storage space and a small initial map which grows as the robots explores its environment. Localization has not yet been attempted using Dynamically Expanding Occupancy Grids and a Centralised Storage System. This research is geared towards implementing Monte-Carlo Localization methods (Fox, Burgard, Dellaert, and Thrun 1999; Dellaert, Fox,Burgard, and Thrun ; Thrun, Fox, Burgard, and Dellaert 2001; Fox, Thrun, Burgard, and Dellaert 2001) to robots using Dynamically Expanding Occupancy Grids. By using this approach this research aims to provide a complete mapping and localization implementation for robots using dynamically expanding occupancy grids and a centralized storage system.Item Nonlinear control with two complementary Lyapunov function(2016-12) Zelenak, Andrew J.; Landsberger, Sheldon; Pryor, Mitchell; Deshpande, Ashish; Fernandez, Benito; Kautz, DougIf a Lyapunov function is known, a dynamic system can be stabilized. However, computing or selecting a Lyapunov function is often challenging. This dissertation presents a new approach which eliminates this challenge: a simple control Lyapunov function [CLF] is assumed then the algorithm seeks to reduce the value of the Lyapunov function. If the control effort would have no effect at any iteration, the CLF is switched in an attempt to regain control. There is some flexibility in choosing these two complementary CLF’s but they must satisfy a few characteristics. The method is proven to asymptotically stabilize a wide range of nonlinear systems and was tested on an even broader variety in simulation. It was also tested on an industrial robot to provide compliant behavior. The simulated and hardware demonstrations provide a broad perspective on the algorithm’s usefulness and limitations. In comparison to the ubiquitous PID controller, the algorithm’s advantages include enhanced performance, ease of tuning, and extensions to higher-order and/or coupled systems. Those claimed advantages are validated by a test with four engineering students, which validates the controller as a viable option for nonlinear control (even at the undergraduate level). The algorithm’s drawbacks include the necessity of a dynamic model and, when linearization is required, the reliance on a small simulation time step; however, for the motivating application –interactive industrial robotic systems – both requirements were already met. Finally, the developed software was released to the public as part of the Robot Operating System (ROS) and the details of that release are included in this report.Item Planar Multicontact Locomotion Using Hybrid Zero Dynamics(2013-11-26) Lack, Jordan ThomasThis thesis proposes a method for generating multi-contact, humanlike locomotion via a human-inspired optimization. The chief objective of this work is to offer an initial solution for obtaining multi-domain walking gaits containing domains with differing degrees of actuation. Motivated by the fact that locomotion inherently includes impacts, a hybrid systems approach is used. Through Lagrangian mechanics, a dynamic model of the system is derived that governs the continuous dynamics, while the dynamics during the impacts are modeled assuming perfectly plastic impacts in which the ground imparts an impulsive force on the impacting link. Using the dynamic model of the planar bipedal robot Amber 2, a seven link biped, a human-inspired optimization is presented which leverages the concept of zero dynamics, allowing for a low dimensional representation of the full order dynamics. Within the optimization, constraints are constructed based on the interaction be- tween the robot and the walking surface that ensure the optimized gait is physically realizable. Other constraints can be used to influence or ?shape? the optimized walking gait such as kinematic and/or torque constraints. This optimized walking gait is then realized through the method of Input/Output Linearization. Finally, the utilization of online optimization in the form of a quadratic program increase the capabilities of simple Input/Output Linearization by introducing a notion of optimality as well as the ability to distribute torque as necessary to meet actuator requirements. Ultimately the combination of the flexability of the human-inspired optimization along with the controllers described result in not only multi-domain human-like walking, but even more importantly a tool for rapidly designing new walking gaits.Item A security and safety rover made from off-the-shelf parts(2014-12) Wang, James Ho; Aziz, AdnanThis report describes the security and safety rover capable of monitoring the the indoor areas of a home and alerting the homeowner of events, such as burglary, fire, or flood. Physical security has always been a concern for homeowners. Digital video security systems are becoming commonplace in modern homes. The majority of the cameras available in this consumer market are fixed-mount cameras with limited capability. The pan-tilt-zoom feature is much less common due to cost. However, with the recent availability of low-cost, powerful, micro-controller boards along with high definition cameras and a myriad of electronic sensors this nascent product comes to mind. This report describes an experimental effort to build a robot using ``off-the-shelf'' hardware and custom software. Specifically, it will answer the question: ``Is it possible to build such a rover with the requisite capability and, if so, at what cost?''Item Standardization for intelligent detection and autonomous operation of non-structured hardware, and its application on railcar brake release operation(2015-05) Hammel, Christopher Scott; Tesar, Delbert; Ashok, PradeepkumarThis thesis introduces a standard framework for evaluating and planning for desired autonomous (or semi-autonomous) operations, then applies the framework, in detail, to the task of automating emergency brake release before rail-car decoupling. A significant hurdle to be accounted for is the lack of standardization of much of the hardware of interest in industry. Non-standardized rail car components must be formally structured as fully as possible to improve the reliability of the robotic automation. This brake release task requires either pushing or pulling a “bleed rod” that protrudes from the side of each rail car. The requirements for each step of the evaluation and planning process will be laid out in this thesis, as an example of how it should be applied to future automation tasks.