Ranjan Mukherjee - US grants

This research project deals with the design and development of a unique spherical mobile robot having a spherical exo- skeleton, an internal mechanism for self-propulsion, and a variety of sensors for motion control and reconnaissance. Unique Features of the robotic system include its retractable camera, retractable manipulators, and telescopic limbs. The arms and limbs are deployed for manipulation and support when the robot is at rest and retracted before the robot resumes its motion. The robot is able to perform rapid maneuvers and move over rough terrain with relative ease. In the process of development of the spherical robotic system, three fundamental problems in the areas of dynamics, control, and design of mechanical systems are addressed. The first problem relates to the development of a nonlinear feedback control strategy for nonholonimic systems with primary application to the rolling sphere. A smooth and time- invariant controller will be developed for the reconfiguration of the sphere. Despite significant progress in nonholonomic control systems, the reconfiguration of the rolling sphere is still an open problem. The second problem of this research is to design an internal mechanism for self- propulsion of the sphere and to construct a closed loop controller for the effective operation of the mechanism. The mechanism provides the sphere with the capability to accelerate, move with constant velocity, or servo at a point. The final problem relates to the development of the spherical mobile robot, which is expected to achieve autonomy thorough coordination between sensing and control. This problem deals with design of the truss, choice and placement of sensors, and design f the overall control system through integration of the two controllers mentioned above. The nonholonomic control strategy, developed as a part of this research, is incorporated in the Advanced Control Systems course, offered in the Department of Mechanical Engineering at Michigan State University. The internal mechanism for self-propulsion of the sphere and the design of the spherical mobile robot are used as motivational examples in a number of graduate and undergraduate courses in design, structures, mechanics, and controls.

DESCRIPTION (provided by applicant): The long term objectives of our research is to test the following two hypotheses: (i) A robotic device with haptic and ultrasound capabilities can accurately examine the human breast, and (ii) A physician can remotely and accurately examine the human breast using such a robotic device. Ultimately, such a robotic device could be used to do screening or focused breast exams for patients in remote areas without access to physicians. If the device is successfully validated, it could also be used to train healthcare professionals in breast pathology, including cancer. Because of incorporation of ultrasound capabilities, examinations by the robotic device might prove to be more accurate than examination by the physician's own hand. In order to achieve these long term objectives, we propose to develop a haptic interface for telediagnostics of breast pathology. These system will be comprised of a robotic manipulator equipped with tactile sensing and ultrasonic imaging capabilities, physician interface capable of rendering haptic information, and two-way audio and video for remote presence capability, all of which will be integrated over the Internet. This research is proposed by a multidisciplinary team comprised of physicians, mechanical engineers, electrical engineers, and computer scientists. It builds upon recent advances in robotics, sensor technology, networking, control systems technology and diagnostics of breast pathology. Upon successful completion of this research, we will be uniquely poised to achieve our long-term objectives, which will in turn improve health care quality for under served communities.

Enhancing Controllability and Observability in Under-Actuated/Under-Sensed Systems through Switching: Application to Vibration Control

Abstract

Many dynamical systems employ transducers that can function both as actuators and sensors and the objective of this research is to investigate under actuation and under-sensing in such systems for reduction in control system hardware. As compared to the traditional approach, where transducers are used as dedicated actuators and sensors, we propose to continuously switch the functionality of the transducer elements between actuator and sensor modes such that each element effectively serves both as an actuator and sensor. This provides the scope for significant reduction in the number of transducers and associated hardware without any loss in controllability or observability. We propose to develop switching, control, and sensing algorithms for optimal system performance, but since these algorithms will depend on the transducer type we will focus on the specific problem of vibration suppression using capacitive piezoelectric transducers. In particular, we will (a) determine the optimal number and location of transducers in a given structure, wherein each transducer is used both as an actuator and sensor, but not simultaneously, (b) determine the "best" fixed partition of the transducers into actuators and sensors, knowing that their roles will reverse with every switching, (c) design a rule for partitioning the transducers into actuators and sensors at every switching based on feedback, rather than using a fixed partition, (d) determine optimal switching intervals, optimal controller and observer gains, and optimal number of switchings, (e) investigate the possibility of fast switching with the objective of deriving the benefits of collocation without sacrificing stability, and (f) conduct experiments to ascertain enhancement of controllability and observability, investigate the merit of introducing under-actuation and under-sensing, address the challenges of implementing switching algorithms, and validate the theoretical results. The concept of switching transducers between actuator and sensor modes, although simple, is novel and has not been explored earlier. This research will provide the scope for significant reduction in control system hardware and although it is being proposed here for piezoelectric transducers, it can be extended to other transducer types and profitably applied to many other dynamical systems, such as active magnetic bearings, MEMS resonators, and thermoelectric cooling devices.

OISE-0609094
Mukherjee, Ranjan (Michigan State University), Planning Visit to the University of Tokyo for Collaboration on Humanoid Robotics Research

This proposal is for a planning visit to Japan to strengthen a collaboration with University of Tokyo Professor Yoshihiko Nakamura in the area of humanoid robotics, specifically concentrating on human-like gaits for humanoids. The proposed work adopts a multi-faceted approach drawing on mechanical engineering and kinematic and kinetic analyses to impart new capabilities in humanoids through collective investigation of problems in multiple technical areas. The U.S. research team includes the PI, Professor Gordon J. Alderink of Grand Valley State University, and one graduate mechanical engineering student from Michigan State University. The planning visit will also include visits to humanoid robotics labs and research centers at Waseda University, National Institute of Advanced Industrial Science and Technology (AIST), Advanced Telecommunications Research Institute (ATR), and Sony Corporation.

This award is funded under the American Recovery and Reinvestment Act of 2009 (Public Law 111-5). The objective of this research award is to expand the sphere of application of impulsive control, and importantly, translate impulsive control from theory to practice. This will be achieved through investigation and experimental implementation of impulsive control in under-actuated dynamical systems. The main problems to be addressed are balancing, walking, and stair-climbing in bipeds; but swing-up control of the pendubot and acrobot will be investigated first. These are benchmark problems in under-actuated systems. The motivation for addressing these problems is to compare the results of this research with results in the literature and establish the advantages of impulsive control for these systems; implement impulsive control for these systems in experiments, which are relatively simple in terms of complexity; and understand the effect of impulsive control inputs on system behavior with the objective of learning to deal with impulsive forces arising from intermittent contact in bipedal locomotion and other robotic tasks.

Impulsive forces are not included in the set of admissible controls for most dynamical systems. By establishing the advantages of impulsive control of under-actuated systems, this research will initiate exploration of impulsive control for a wide range of problems such as orbital transfer of satellites, rapid maneuvering of missiles and underwater vehicles, and control of prosthetic devices. This award will have broad impact through integration of research and education, diversity, and outreach. It will provide research topics to graduate students and contribute towards the development of future generation engineers and academics. The research results will be integrated into select courses and published in archival journals and efforts will be made to increase the participation of minorities. The outreach component will include sharing results with researchers at federal government laboratories and collaboration with faculty in other universities.

This research will investigate fundamental problems in fluid-structure interactions motivated by the development of an efficient and maneuverable submersible. This submersible employs a fish-like flexible tail driven by a conveyed fluid jet; the combination of the jet's thrust and that of the oscillating tail synergistically propels and maneuvers the submersible. Two disconnected bodies of fluid-structure interaction literature will be merged in this research - the study of oscillatory fish-like propulsion and the study of fluttering fluid-conveying pipes. The fluid-conveying pipe literature will be extended toward conditions that exist in fish-like swimming: external flow, spatially-variable tail planforms, large deformations, and fast accelerations and large rotations. Furthermore, acceleration and rotation of the submersible will be accomplished by varying the conveyed fluid velocity; an area, which requires additional theoretical development. Euler-Bernoulli beam theory and the theory of elastica will be used to model spatially-varying beams conveying fluid with time-dependent velocity. These theoretical investigations will be complemented by numerical simulations and a hardware platform for experimental verification.

This fish-like submersible uses a novel mechanism for propulsion which requires investigation of a flexible tail subjected to significant accelerations and fluid forces. The submersible has many potential applications; since the surrounding water is pulled into the hull during normal operation, it is an ideal candidate for environmental monitoring, chemical sensing and water cleanup operations. The methods developed in this research will also apply to other emergent problems in fluid structure interactions, such as vibration control of large wind turbine blades and better design of ornithopters through improved understanding of bat and insect flight. In humans, deflection of a flexible member in the airway causes snoring and sleep apnea - better understanding of fluid loading on flexible structures could lead to improved therapy and surgical treatments.

In a fully actuated mechanical system, an independent control force is available for each degree of freedom. Systems with fewer independent controls than degrees-of-freedom are called "underactuated." Many important systems are naturally of this class, including certain types of missiles, satellites, underwater vehicles, and bipedal robots. As might be expected from the shortage of control inputs, underactuated systems are challenging to stabilize and to steer. This project improves upon existing approaches by using large-amplitude, short-duration control forces -- called impulsive forces -- to expand the region of attraction of the system. The region of attraction is the set of initial configurations and velocities from which the system will be able to reach its desired operating point. If the system ever leaves its region of attraction, it will depart from its desired behavior, corresponding to, for example, the loss of control of a missile or spacecraft. Expanding the region of attraction means safer operation, since the system can tolerate upsets due to large disturbance forces or unexpected initial conditions. A major new contribution of this project will be to bridge the gap between purely theoretical studies and practical applications, through careful experimental validation on widely accepted benchmark test bed systems.

Dynamical systems are continuously subjected to disturbances and their ability to reject them largely depend on the stability property of their equilibrium configurations. This research will enlarge the region of stable operation of underactuated dynamical systems and reduce the incidence of unstable behavior, behavior that typically has negative consequences. The region of stable operation will be enlarged by application of impulsive forces, which will be included in the set of admissible control inputs. In this project, impulsive control will be implemented using high-gain feedback over short intervals of time and mathematical tools such as singular perturbation methods will be used to analyze and design the high-gain feedback systems. The implementation of high-gain feedback will require the use of very fast observers to estimate the unmeasured velocities. To this end, high-gain observers will be designed and analyzed via multi-time-scale singular perturbation methods. There are many challenging technical issues that need to be solved to make such designs feasible. This includes the development of efficient algorithms that can transfer the system configuration from outside the region of stable operation to inside the region, using a single application or multiple applications of impulsive inputs. The developed algorithms will be tested experimentally on simple underactuated systems; but the methods will be applicable to more complex problems such as orbital transfer of satellites, rapid maneuvering of missiles and underwater vehicles, and active prosthetic devices.

Children with severe movement impairments often depend on assistive devices such as wheelchairs, robotic arms, or communication aids for activities of their daily lives. However, learning to control these complex devices can be challenging for children. In this project, the investigators explore the development of a general purpose body-machine interface that allows children to control several devices using movements of the body. A key advantage of this type of interface is that it is non-invasive, easy to wear, and can "grow" with the child. The project will enhance basic understanding of how body movements can be exploited to control external devices, especially in children. The project will have an impact on the independence and quality of life for children with severe movement impairments. A key challenge of this research is to translate the body movements of the user into commands that can control the robotic arm for performing an intended task. The traditional approach to this translation problem has been based on principal components analysis (PCA); however, this is not well suited for complex tasks requiring the control of many degrees of freedom. In this project, the investigators will develop alternative, novel methods that can take full advantage of the movement repertoire of the user and can also yield intuitive control. In addition to the significant impact of this research on the quality of life of individuals with disabilities, the project includes outreach to K12 students and teachers to interest them in STEM and improve their understanding of disabilities and motor development.

The current work will be carried out through three aims. These are: 1) utilize a principal component analysis-based approach for control of high degree of freedom movements within the robot; 2) develop a virtual body model approach; and 3) determine learning characteristics of high-degree of freedom body machine interfaces in children. Each aspect of the research will be assessed through rigorous performance metrics. Specifically, the objectives of this project are to: capture a wide range of motion patterns using sensors on the head and upper body; and to map anatomically distinct motion patterns to commands for controlling the robot arm using a virtual model of the user's body. Based on the sensor data from the user, algorithms will be used to calculate how the virtual model bends, twists, and turns; and these deformations will then be translated into commands that control the robotic arm. In addition to developing these methods, the investigators will also evaluate the intuitiveness of these new methods by testing how quickly children can learn to control a robotic arm using these methods.