F.G. Pin’s research while affiliated with Oak Ridge National Laboratory and other places

Oak Ridge National Laboratory, as well as the rest of the U.S. Department of Energy community, has numerous off-gas stacks that need to be decommissioned, demolished, and packaged for disposal. Disposal requires a waste disposition determination phase. Process knowledge typically makes a worst-case scenario decision that may place lower-level waste into a more expensive higher-level waste disposal category. Truly useful radiological and chemical sampling can be problematic on old stacks due to their inherent height and access hazards, and many of these stacks have begun to deteriorate structurally. A remote stack characterization system (SCS) that can manage sample and data collection removes people from the hazards and provides an opportunity for access to difficult to reach internal stack areas. The SCS is a remotely operated articulated radiological data recovery system designed to deploy down into off-gas stacks from the top via crane. The battery-powered SCS is designed to stabilize itself against the stack walls and move various data recovery systems into areas of interest on the inner stack walls. Stabilization is provided by a tripod structure; sensors are mounted in a rotatable bipod underneath the tripod. Sensors include a beta/gamma/alpha detector, a removable contaminant multi-sample automated sampler, and a multi-core remote core drill. Multiple cameras provide remote task viewing, support for sampling, and video documentation of the process. A delay in funding has delayed project delivery somewhat. Therefore, this paper describes the technology and shows fabrication and testing progress to the extent that data is available.

In typical teleoperated surgeries, skilled staff are still necessary in the remote surgical room to change manipulator tooling and to manage surgical supply delivery and removal. This paper describes the development of a nurse robot to provide automated support to a teleoperated surgical manipulator system in environments where the presence of skilled surgical support staff may not be practical. The tools must be inserted precisely into a compliant manipulator in a timely manner, and the supplies are diverse in nature. To support experimental investigations and evaluations, a seven degrees-of-freedom commercially available manipulator was selected. The design of novel end-effectors, tool grasping and supply holding features, and tool auto-loading systems for optimum surgical tool changing and supply delivery in minimum time is presented. A novel approach for calibration of the nurse robot among compliant and rigid subsystems and for managing forces during subsystem interaction is described and experimental results using this force management approach are presented. Overall experimental performance data for the nurse robot system during tool changing and supply delivery tasks is also presented to illustrate the feasibility of performing these functions in a remote medical or trauma care-assist cell.

To support an Oak Ridge National Laboratory programs exoskeleton project, a unique multi-axis foot force/torque sensor was constructed and tested that has biomedical application such as clinical gait analysis. The challenging aspect of this multi-axis force sensor is that it had to conform to the bending of the human foot, withstand high impact loads, have high force sensitivity, feel comfortable to the human wearer, be integrated into a military style boot, measure the forces on the human foot when either the ball or the heel of the foot is in contact with the ground, respond to both positive and negative loading and have a low overall height and weight. This paper describes the design and testing of this unique sensor.

Most hydraulic servo systems are designed with little consideration for energy efficiency. Pumps are selected based upon required peak power demands, valves are chosen primarily for their rated flow, actuators for the maximum force. However, the design of a hydraulic servo system has great potential in terms of energy efficiency that has, for the most part, been ignored. This paper describes the design and control of a large-scale ship motion simulation platform that was designed and built at Oak Ridge National Laboratory for the Office of Naval Research. The primary reasons to incorporate energy-efficiency features into the design are cost and size reduction. A preliminary survey of proposed designs based on traditional motion simulation platform configurations (Stuart Platforms) required hydraulic power supplies approaching 1.22 MW. This manuscript describes the combined design and control effort that led to a system with the same performance requirements, however requiring a primary power supply that was less than 112 kW. The objective of this paper is to illustrate alternative design and control approaches that can significantly reduce the power requirements of hydraulic systems and improve the overall energy-efficiency of large-scale hydraulically actuated systems.

Sodium borohydride's properties make it a good source of hydrogen for use with a fuel cell for an on-demand system that is easily controllable and has no idle costs. Previous work, as described in the literature, indicated that ruthenium (Ru) is an efficient catalyst for generating hydrogen from sodium borohydride. Tests were conducted to evaluate catalyst loading with the results of these tests indicating that the hydrolysis rate is affected by the loading of the catalyst. It was also apparent that the substrate surface is not completely occupied by Ru at the lower loadings, and that increased loadings are needed to optimize the reaction rate. A differential rate test with a fixed bed reactor was also conducted. It was observed that temperature has a significant effect on the rate of reaction. Feed rate also affected the rate of reaction with lower feed rates (longer residence time in the reactor) having higher reaction rates. A bench-top hybrid system was also developed and tested. This test bed demonstrated how a system based on a chemically generated hydrogen-fed proton exchange membrane fuel cell could be integrated with batteries to provide a hybrid power system that can meet the demands of a highly varying electrical load up to four times the rated output of the fuel cell.

The decrease in manpower and increase in material handling needs on many naval vessels provides the motivation to explore the modeling and control of naval robotic and robotic assistive devices. This paper presents a simple methodology to symbolically compute the dynamic equations of motion of a serial link manipulation system operating on the moving deck of a ship. First we provide background information that quantifies the motion of the ship, both in terms of frequency and amplitude. We then formulate the motion of the ship in terms of homogeneous transforms. Likewise, the kinematics of a manipulator is considered as a serial extension of the ship motion. We then show how to use these transforms to formulate the kinetic and potential energy of the arm moving on a ship. As a demonstration, we consider two examples: a one degree-of-freedom system experiencing three sea states operating in a plane to verify the methodology and a 3 degree of freedom system experiencing all six degrees of ship motion to illustrate the ease of computation and complexity of the solution. We provide a preliminary comparison between conventional linear control and repetitive learning control (RLC) and show how fixed time delay RLC breaks down due to the varying wave disturbance frequency.

Unlike traditional assembly line robotic systems that have a fixed kinematic structure associated with a single tool for a structured task, next-generation robotic surgical assist systems will be required to use an array of end-effector tools. Once a robot is connected with a tool, the kinematic equations of motion are altered. Given the need to accommodate evolving surgical challenges and to alleviate the restrictions imposed by the confined minimally invasive environment, new surgical tools may resemble small flexible snakes rather than rigid, cable driven instruments. Connecting to these developing articulated tools will significantly alter the overall kinematic structure of a robotic system. In this paper we present a technique for real-time automated generation and evaluation of manipulator kinematic equations that exhibits the combined advantages of existing methods-speed and flexibility to kinematic change--without their disadvantages.

During robot path planning and control the equations that describe the robot motions are determined and solved. Historically these expressions were derived analytically off-line. For robots that must adapt to their environment or perform a wide range of tasks, a way is needed to rapidly re-derive these expressions to take into account the robot kinematic changes, such as when a tool is added to the end-effector. The JFKengine software was developed to automatically produce the expressions representing the manipulator arm motion, including the manipulator arm Jacobian and the forward kinematic expressions. Its programming interface can be used in conjunction with robot simulation software or with robot control software. Thus, it helps to automate the process of configuration changes for serial robot manipulators. If the manipulator undergoes a geometric change, such as tool acquisition, then JFKengine can be invoked again from the control or simulation software, passing it parameters for the new arm configuration. This report describes the automated processes that are implemented by JFKengine to derive the kinematic equations and the programming interface by which it is invoked. Then it discusses the tree data structure that was chosen to store the expressions, followed by several examples of portions of expressions as represented in the tree. The C++ classes and their methods that implement the expression differentiation and evaluation operations are described. The algorithms used to construct the Jacobian and forward kinematic equations using these basic building blocks are then illustrated. The activity described in this report is part of a larger project entitled ''Multi-Optimization Criteria-Based Robot Behavioral Adaptability and Motion Planning'' that focuses on the development of a methodology for the generalized resolution of robot motion equations with time-varying configurations, constraints, and task objective criteria. A specific goal of this project is the implementation of this generalized methodology in a single general code that would be applicable to the motion planning of a wide class of systems and would automate many of the processes involved in developing and solving the motion planning and controls equations. This project is funded by the U.S. Department of Energy's Environmental Management Science Program (DOE-EMSP) as project EMSP no. 82794 and is transitioning to the DOE-Office of Biological and Environmental Research (OBER) as per FY-02.

Navigation of autonomous mobile robots in unknown and unpredictable environments is a challenging domain to test and/or demonstrate knowledge representation and reasoning techniques because it involves a number of unique characteristics: The input to the control system, particularly when provided by sonar range finders and odometric wheel encoders, is inaccurate, sparse, uncertain, and/or unreliable. No complete mathematical representation exists of the process termed “navigation,” although, as demonstrated by humans, a set of skills for accomplishing this process exists that can typically be represented in a linguistic manner as IF-THEN rules (e.g., if the goal is to the left, then turn left; if an obstacle is detected to the right, then bear left). The approximations involved in the numerical representation of the system and its environment (e.g., geometric representations, map discretization in grid) are significant. A navigation environment is in general dynamic and unpredictable, typically leading to large uncertainties in its representation.

Most fuzzy logic-based reasoning schemes developed for robot control are fully reflexive, i.e., the reasoning modules consist of fuzzy rule bases that represent direct mappings from the stimuli provided by the perception systems to the responses implemented by the motion controllers. Due to their totally reflexive nature, such reasoning systems can encounter problems such as infinite loops and limit cycles. In this paper, we propose an approach to remedy these problems by adding a memory and memory-related behaviors to basic reflexive systems. Three major types of memory behaviors are addressed: memory feeding, memory management, and memory utilization. The concepts underlying these major classes of behaviors are first presented. An example of their implementation is then described for the recognition of limit cycles during the sensor-based navigation of mobile robots. Results of sample cases involving an autonomous robot navigating in a priori unknown environments are presented and discussed to illustrate the efficiency of the approach.