Ralph L. Hollis - US grants

9503992 Hollis The objectives of this project are: (1) providing the postdoctoral research associate with a rich environment for extending previous work in distributed real-time control of dynamic systems to the problem domain principally of manufacturing automation; (2) to contribute in the area of distributed control to the developing project in miniature factories, and (3) to provide a means for the associate to broaden the exposure to a wide variety of research topics within several units of the university. Given the relatively constant cost of mechanical actuators and mechanisms in contrast to the continuing advances in price, size, and performance of computers it is inevitable that computational processes become as pervasive and distributed as their mechanical counterparts. The intent of this research project is to `egin exploration of scientifically grounded approaches to the construction of such systems. Specifically, distributed real-time control systems will be constructed primarily to support a highly modular (in mechanical hardware, computational hardware, and software) miniature assembly system. The proposed controller architecture makes use of an event driven model for parallel processing to afford the necessarily high levels of software and hardware modularity, both of which will allow for rapid and efficient system development and reconfiguration without risking catastrophic failure due to unforeseen computational hardware or software interactions. ***

9527190 Hollis This research aims to improve assembly in product manufacturing. Assembly is a difficult and time-consuming process to automate, especially when small tolerances are necessary. This architecture for agile assembly is a strategic framework using agent-based robotic elements that are precise, modular, and extensible. These elements form sets of leased self-contained software/hardware modules that can be programmed and operated over the Internet, and can be brought together to form miniature agile factories for assembly. Such a scheme requires a combination of intelligent networked communication, distributed computing resources using high-performance processors, and distributed sensor/actuator subsystems of novel design. Design and construction of a prototype miniature factory according to the principles of agile assembly architecture will be carried out to validate the research and provide a unique and powerful reconfigurable platform for assembly research and evaluation by industry. Partial assembly of magnetic storage disk drives will serve as a testbed. This work aims to advance knowledge in three main areas: (1) seamless wide area and local area networking of distributed agents emphasizing high quality of service; (2) modular precision robotic assembly elements containing high-performance embedded processors; and (3) a comprehensive software environment for modeling, simulation, and programming of the miniature factories. The results of the research could allow manufacturers to develop a capability for geographically distributed design and deployment of assembly systems while providing drastically reduced changeover times, higher quality, and a new level of manufacturing system portability.

This project advances knowledge about how to give computer users convincingly real haptic (sense of touch) interaction with computers. While there has been some progress in this area, chiefly through the use of back-driven robotic-like manipulators, this is a substantially new approach which promises a qualitative leap in improvement of such capabilities: A user interacts with the computer by grasping arigid tool whose behavioral description is computed, employing this tool to interact with computed environments which are semantically meaningful in terms of the application. At the same time, the environment exerts realistic forces and torques on the tool's handle which are felt by the user. The vision is one of providing the computer user immediate, high-fidelity, convincingly real interaction with computed environments. The new haptic interface approach is based on a recently developed magnetic levitation technology, and on recent strong advances in the art of physically-based simulation. The magnetic levitation technology uses Lorentz forces to stably levitate and control a rigid body (which includes the handle through which the user interacts) in six degrees of freedom, giving a new and heretofore unexplored physical basis for haptic interaction. The new simulation methods are based on special-purpose techniques that promise to be simpler, faster, and more robust than their generic counterparts. Robust realistic physical simulation, coupled with high- bandwidth six-degree-of-freedom force reflection, has the potential to greatly improve the state of the art in the feel, performance, and capabilities of virtual environment systems for use in a wide range of human activities.

9523156 Hollis Planar linear motors, often called forcers, provide low friction motion in a large workspace with few moving parts. Because they operate as open loop steppers, they are: susceptible to losing steps, affected by external disturbances, and cannot provide controlled forces or high stiffness. Feedback control can compensate for these deficiencies if reliable and unobtrusive position sensors are available. This project will investigate two unobtrusive sensing techniques for planar motor systems with 1 micrometer positioning resolution. One proposed sensor uses the AC magnetic field drive flux to induce a field in a sensing tooth on the mounting platen. The other approach also uses platen teeth, but is based on optics. A when a tooth coated with a florescent dye material is illuminated by light of one wavelength, it emits light at a longer wavelength; the intensity of this secondary wavelength is function of position. Research issues for both sensors are spatial and temporal noise, cross talk, and sensor-forcer integration. Using these sensors, a model based and a variable structure robust control scheme will be designed to deal with nonlinear magnetic fields, payload effects and model uncertainties. This research will enable increases in speed and payload for planar linear motors, making them attractive as actuators for next generation machines and equipment that need one micrometer positioning over a large planar workspace. ***

*** 9802191 Hollis This project builds on results from previous NSF grant IRI9420869: "Magnetic Levitation Haptic Interfaces." The previous grant developed a haptic interface (force/torque feedback) interaction capability for computer users based on Lorentz magnetic levitation. The user interacts with a single moving part with three rotational and three translational degrees of freedom (DOF) floating above the desktop. Combined with advances in physically-based simulation methods, it provided computer users convincingly real haptic (sense of touch) interaction with computers through six-degree-of-freedom (DOF) position input and 6-DOF force/torque output to the user's hand at resolutions of approximately 10 microns and a position bandwidth of approximately 75 Hz. The present effort seeks to make several technology enhancements and to quantitatively measure the psychophysical effectiveness of this approach. Comparisons are being made for haptic interaction with virtual environments and telemanipulation environments, and both of these with real environments to quantitatively clarify the differences between these modes. Methods for robustly "caching" local haptically-relevant regions of the virtual environment are being explored to smoothly synchronize the visual and haptic displays. Psychophysical methods are employed in the design of the human-computer interface to provide a controlled setting for conducting psychophysical measurements. Finally, detailed psychophysical measurements are being performed with the developed experimental system to quantitatively measure the degree of reality provided. The knowledge gained from this work provides needed information for the engineering science of haptic interface design while helping to elucidate the nature of haptic interaction itself. ***

This project will investigate the set of problems which must be solved to produce a new class of precision motion machines capable of rapidly moving over a planar surface and subsequently positioning to sub-micrometer precision without the need to provide air, power, and signals over a tether as is the case with today's technology. Planar linear motors currently used in industry operate with a tether in an open-loop stepping manner. This mode of operation makes them restricted in range of motion, susceptible to loss of steps, unable to reject external disturbances (such as tugs on the tether), unable to provide controlled forces, and unable to provide high stiffness. A previous NSF-funded project produced technology advances allowing integrated robust position sensing and control for this class of motors, providing improved motion characteristics including better disturbance rejection, better resolution, and reduced energy consumption. Building on these results, the present project will explore optimal energy utilization, innovative frictionless bearing design, and innovative integrated computing and communications technology to eliminate the need for a tether.
Elimination of the tether will allow machines based on planar motors to be completely free-roaming over their corresponding stator surfaces. This will enable significant improvement in the operation of a wide range of systems, including manufacturing systems, that are based on these motors.

This project, developing, replicating, distributing, and using a new magnetic levitation haptic interface system, provides direct electrodynamic interaction with a single moving part permitting high-fidelity six-degree-of-freedom (6-DOF) sensing and force/torque capability (unlike currently available haptic systems which resemble small, back-driven robot arms with motors, encoders, pivots, links, and transmission elements). Haptic interfaces allow computer users to interact mechanically, as well as visually, with computed information. With the new method, the user grasps the freely-levitated handle (manipulandum) of a desktop-high device, maneuvering it in 6-DOFs to provide position and force/torque information to a physically-based 3D simulated environment with gravity, hard contact, flexible deformation, friction, and texture attributes. The running simulation provides 6-DOF force/torque output to the manipulandum, and consequently to the hand. Both the proprioceptive (kinesthetic) senses of the fingers, hand, and wrist as well as the tactile senses in the skin are involved in the interaction. The prototype magnetic levitation haptic system provides higher bandwidths and resolutions than other existing techniques-an important consideration for conveying subtle friction and texture information to the user. Dramatically reducing cost, this project will greatly improve the performance of the current prototype system, replicating eight new systems using state-of-the-art manufacturing techniques. The developed systems will be distributed to seven haptic researchers in the nation. The new systems provide a basis for supporting seven new independent research efforts. Thus, eight projects, involving six universities, form part of the project.

Hollerback, Utah: investigating and displaying manual control dynamics, assisting in determining how humans interact to assemble, disassemble, and manipulate CAD objects
Hollis, CMU: understanding two-handed manipulation and investigating whether blind persons can benefit from haptic communication
Howe, Harvard: cost-benefit tradeoff for haptics as a function of frequency response, providing insights into fundamental tactile perception and motor control mechanisms and guidelines for cost-effective haptic interfaces
James, CMU: new deformable rendering algorithms for a large class of flexible models enabling technology for computer graphics and virtual environment simulation
Khatib, Stanford: new control algorithms for haptic display enable new desktop haptic applications
Pai, Rutgers: new kinds of audio-haptic interfaces with tightly synchronized sounds and contact forces enhancing understanding of human perception of contact rendered using a haptic interface
Tan, Purdue: understanding perceptual dimensionality, texture perception, & multimodal rendering of information

Many of the research efforts involve undergraduate students, some from under-represented groups. High-fidelity haptics has enormous potential for K-12 education.

Robotics and Human Augmentation Program

ABSTRACT


Proposal #: 308067
Title: Toward Dynamically Stable Mobile Robots in Human Environments
PI: Hollis, Ralph
Carnegie Mellon U.


The goal of the research is to gain deeper understanding of how dynamic agility can be achieved in mobile machines interacting with people and operating in normal home and workplace environments. The work develops and utilizes a novel dynamically stable rolling machine research platform. To illuminate many of the issues surrounding the operation of agile machines in human environments, in particular the ability to traverse narrow, cluttered rooms and to function robustly in the presence of people, we are investigating what could be one of the simplest possible dynamically-stable mobile robots: a machine whose body locomotes using only a single spherical wheel. This enables the machine to be tall and narrow, have a high center of mass, respond compliantly to nudges and shoves, and rapidly move in any direction. The machine will have a vestibular system based on an inertial measuring unit and other sensors with an on-board computer to provide closed-loop control of movement and balance. The efficacy of this new type of rolling locomotion will be evaluated in the context of human environments. Insights will be gained toward the development of agile motive platforms that in the future could be combined with the research community's ongoing work in perception, navigation, and cognition, to yield truly capable intelligent mobile robots for use in physical contact with people. Such robots, if realizable and economically viable, would function as aids to elderly or disabled persons; provide guidance and assistance in public spaces; help with education and entertainment; perform domestic cleaning and housekeeping; or fetch and carry everyday objects. Many other uses such as entry into hostile environments, rescue, and automated surveillance to safeguard people or property can be envisioned.

The project concerns high-performance haptic (sense of touch) interaction with three-dimensional (3D) computed (virtual) and real environments. The principal research objective is to quantitatively determine how much reality is achievable in 3D haptic/visual virtual and remote real environments. The approach is based on six-degree-of-freedom (6-DOF) haptic interaction technology using Lorentz magnetic levitation. Both proprioceptive (kinesthetic) senses of the fingers, hand, and wrist as well as the tactile senses in the skin are involved in the interaction. Lorentz levitation provides higher bandwidths and motion resolutions than are available with traditional technologies. The project includes i) adding direct force-torque sensing, combining favorable aspects of both impedance- and admittance-type devices; ii) the creation of a highly realistic 3D peg-in-hole virtual environment; iii) development of an elastic deformation environment with buckling phenomena; iv) comparison of subjects' interaction with virtual, real, and remote-real environments, and v) performance of a suite of psychophysical experiments to quantitatively measure the degree of reality provided. The quantitative characterization of haptic interaction transparency afforded by this approach contrasts markedly with pure engineering measurements or purely subjective evaluations. The research results provide knowledge for the engineering science of haptic interface design while helping to elucidate the nature of the human haptic interaction process. This could lead to the future widespread use of haptic technology for computer augmented design, medical training, telemanipulation and telepresence systems, vehicle piloting simulation, and the exploration of complex multi-dimensional data sets. The project has an important educational impact which includes the study of haptics in undergraduate and graduate course work.

The proposed research builds on results obtained from NSF grant IIS-0308067. The goal of the research is to gain deeper understanding of how dynamic agility can be achieved in mobile machines interacting with people and operating in normal home and workplace environments. The work utilizes and further develops a novel dynamically stable rolling machine research platform referred to as a "ballbot." In the spirit of minimalism, the machine balances and locomotes using only a single spherical wheel. This enables the machine to be tall and narrow, have a high center of mass, respond compliantly to nudges and shoves, and rapidly move in any direction. The purpose is to illuminate many of the issues surrounding the operation of agile machines in human environments; in particular the ability to traverse narrow, cluttered rooms and to function robustly in the presence of people. A key aspect of the project is to investigate the addition of a pair of dynamically significant two-degree-of-freedom arms. This will allow the development of control strategies for movement of the ballbot body by dynamic swinging of the arms. Automatic recovery from unplanned collisions with a spectrum of obstacle types as well as purposeful manipulation of the human environment, e.g., carrying unknown loads while balancing, and opening/closing drawers and doors will be investigated using only intrinsic sensing abilities. Insights will be gained toward the development of agile motive platforms that can be combined with the research community's ongoing work in perception, navigation, and cognition, to yield truly capable intelligent mobile robots for use in physical contact with people. Such robots, if realizable and economically viable, would function as aids to elderly or disabled persons; provide guidance and assistance in public spaces; help with education and entertainment; perform domestic cleaning and housekeeping; or fetch and carry everyday objects

This project is creating a novel modular, distributed, and highly scalable computational/physical system for multi-disciplinary research and applications in micro-manufacturing. The approach is a robotic-agent-based distributed information architecture of a type not found in industry today. The distributed nature of the agent-based system requires neither centralized control nor a centralized database for its operation, thereby avoiding communication and code complexity bottlenecks. The goals are to 1) dramatically reduce design, program, and deployment times compared with state-of-the-art systems, 2) greatly increase mechanical precision over existing methods, and 3) greatly reduce floor space requirements.

Intellectual Merit

The architecture is highly generic, and is applicable to multi-step flow-through production systems for domains such as micro-fabrication of small parts, assembly of meso- and micro-scale products, synthesis of discrete amounts of chemicals, and analysis of biological materials. The system addresses several difficult human-computer interface issues making the system more accessible to researchers and students and much easier to use compared with conventional approaches. In this work, collections of computational/physical agents are designed and programmed through a virtual 3D representation that is registered in space and synchronized in time with the actual micromanufacturing system. The system includes several automatic procedures such as calibration and multi-agent coordination which reduces programming tedium. We expect the developed system to be a suitable research platform where multiple, concurrent experiments can be run by faculty and students. The intention is also to provide a working system which can serve as an example for industry. The team has great expertise in this area, and their research will benefit both the research community through the development of new manufacturing techniques, but will also contribute to the field by expanding its scope of applicability.

Broader Impacts

This project provides a vision of compilable factories, and if broadly realized, represents a transformative breakthrough in automated assembly. This includes automated assembly of optical systems, mass spectrometers, 3D devices formed from metal parts, and mass replication of intelligent micro-robots for environmental monitoring. Moreover, education is a vital component of the proposed project. The enhanced minifactory provides a new learning environment for multi-robot programming and cooperative robotics. There is also a close interaction closely with manufacturing engineering faculty and students at Walla Walla University in Washington State. We also continue the fruitful
student exchanges with micro-manufacturing laboratories at Technical University of Munich and the Swiss Federal Institute of Technology. As always, undergraduate students engage in the research and add a valuable and vital contribution to the work.

The project studies three important domains which highlight physical Human-Robot Interaction involved in assisting humans in the home and workplace: (1) human guidance through cluttered environments using physical contact, (2) cooperative carrying of large objects through complex and dynamic environments, and (3) robot assisted sitting and getting up. This is accomplished through the evolution of an experimental single-wheel mobility platform into an autonomous robot that is instructed and guided by people in a natural way.

Such systems are needed in many public health domains, including care for the elderly, rehabilitation and assistive programs. Project results are incorporated into coursework offered by two different departments at CMU, and exposes students to unique robot planning, control and Human-Robot Interaction issues. High school students are introduced to this area as part of the Andrew Leap program. Students from underrepresented groups participate through the ARTSI program. Results may be presented in cooperation with museums or entertainment companies.

Research Objectives and Approaches:

The objective of this research is to create the first spherical induction motor with unlimited rotation and to characterize its performance, principally for use as a spherical robot wheel. The outcome offers reduced mechanical complexity and higher performance when compared with traditional mechanical methods. The approach concentrates on the inductance principle, depending heavily on material characterization, 3D magnetic simulation, solid modeling, electronic circuit design, and real-time software development, keeping in mind the primary application to dynamically stable single-wheel robots.

Intellectual Merit:

The research develops a spherical induction motor permitting direct torque generation and rotation about arbitrary axes with a minimum of mechanical friction or backlash. The results provide an exemplar for a new class of motor, permitting quantitative comparisons with more traditional solutions. The successful outcome simplifies mechanics, improves performance, and lowers the cost of dynamically stable single-wheel mobile robots.

Broader Impacts:

The research investigates important issues in the technology of direct drive spherical actuation which are likely to be generally applicable. The research enhances what is already a new and different approach to mobile robot locomotion, furthering the strong public interest in dynamically stable mobile robots around the world. Besides research experiences for students working on the project, course work offered by two different departments at Carnegie Mellon University is enhanced, exposing students to issues of electromagnetic design, manufacturing, and characterization. The outreach activities expose high school students to hands-on research. There is strong international collaboration with Tohoku Gakuin University in Japan.

The research objective of this award is to develop motion planners and controllers that exploit the dynamics of balancing mobile robots to produce fast, dynamic, graceful motions. The approach has two main components: the development of elemental closed-loop motion policies and the development of a hybrid control architecture that pieces the motion policies together to produce more complex behaviors. Motion policies that respect the underlying dynamic constraints of the system will be created by first planning trajectories in shape space and then creating feedback controllers that track the resulting trajectories. A planning layer that has an understanding of which motion policies can be gracefully composed is used to determine motion sequences that avoid obstacles and achieve some global navigation task. These sequences are encoded in a graph, creating a hybrid control architecture that appropriately switches between motion policies to achieve the desired global motion. Deliverables include algorithm development, software, research documentation, graduate student education, and demonstration of the resulting integrated system on the human-sized ballbot balancing robot. The award will provide opportunities for high- and middle-school girls to interact with balancing robots.

If successful, this award will increase the ability of balancing robots to operate safely and agilely around humans. Balancing robots have many properties that make them ideal candidates as human companions and assistive robots. Unlike statically-stable mobile robots, balancing robots can simultaneously be tall enough for eye-level interaction with humans and narrow enough to negotiate cluttered environments. Balancing robots react naturally when nudged and pushed around, making them unique platforms for physical interaction with humans. The motion planning and control algorithms developed in this project will provide a means for these robots to move in a natural-looking manner through human environments. Results will be disseminated promote further development of these technologies and to inspire the general public. Graduate and undergraduate students will benefit through direct involvement in the research, and 7-12 grade girls will be engaged in activities to explore the potential of balancing robots.

This EArly-concept Grant for Exploratory Research (EAGER) project addresses strategies and software algorithms for manufacturing utilizing teams of mobile robots. The research involves the creation of new algorithms, simulations, and experimental demonstrations utilizing a previously developed minifactory multi-robot micro-manufacturing system.

The specific problem to be addressed is automated discovery and calibration of assembly manufacturing system layouts prior to the commencement of automated assembly, and re-discovery and calibration as the mix or mobile robots and other components changes over hours or days.

The project develops a test bed for exploring highly agile robot mobility in human environments such as an office or home, while also incorporating a highly dexterous ability for purposefully interacting with features in the environment such as opening doors, moving tables or furniture, and manipulating everyday objects. Mobile robots are poised to take increasingly important roles in our daily lives ranging from transporting materials in warehouses and hospitals, inventory management in retail stores, and applications in direct service to people including support and guidance in public spaces. Wheeled mobile robots currently lack the agility and dexterity needed for these kinds of tasks.

The project enhances the CMU ballbot research platform, a first-of-its-kind capability previously developed under NSF funding. Ballbots are dynamically stable mobile robots, which balance and locomote on single spherical wheels. Unique advantages of the ballbot class of mobile robots include tall heights enabling human-robot interaction at eye level, a small footprint afforded by the single ball wheel, and omni-directional motion with intrinsic compliance. This combination of desirable features does not exist in traditional statically stable mobile robots. The project develops and adds a pair of 7-degree-of-freedom arms and hands to the CMU ballbot platform. The objective is to create an unprecedented capability for agile and dexterous mobile manipulation. Here, the term "agile" refers to the ballbot?s current ability to swiftly and smoothly move through the environment, while avoiding static and dynamic obstacles. The term "dexterous" refers to the ability to skillfully manipulate objects and interact with the environment. Reaching this compound goal is enabled in part by the creation of a pair of new lightweight but strong human-scale arms/hands for the ballbot and the requisite software to enable a wide range of future research.