MEMSDUKEPRATT School of engineering

Education

  • PhD, New York University, 1969
  • MS, University of Wisconsin, 1960
  • BE, University of Roorkee, India, 1957
  • B.Sc., Agra University, Agra, India, 1954
Devendra P Garg
  • Office Location: 223 Hudson Hall
  • Office Phone: (919) 660-5330, (919) 660-5296
  • Email Address: dpgarg@duke.edu
  • Professor Garg's areas of interest include modeling, simulation, and control of dynamic systems and robotics. In particular, his research deals with characterization and control of nonlinear phenomena in physical systems such as robots, automated manufacturing, and high-speed ground transportation. In the area of robotics, he is interested in the design of feedback controllers using the concepts of modern control theory. In the area of high-speed ground transportation, he has conducted research on dynamics and control of ground-based, air-cushion and magnetically levitated vehicles.

    One challenging area of research which Professor Garg is currently pursuing deals with the coordination and control of two robots handling a large structural object performing a series of intricate maneuvers in a confined work space. Analytical development of path planning and collision avoidance strategies and their practical implementation are being carried out in the Robotics and Manufacturing Automation (RAMA) laboratory. Two ABB Industrial grade six-degree-of freedom revolute jointed robots are available for these experiments. Major challenges related to this research effort include the identification and design of techniques for incorporating sensory data in the control algorithm, modeling the nonlinear dynamics of the manipulators, and the development of intelligent and adaptive control schemes for the coordination of multiple arms in the presence of unknown parameters and payload variations. Another area of research that Professor Garg is actively pursuing is sensor modeling, data acquisition and management, and data fusion in the context of swarm robotics. For the experimental work in this area, the Robot Control Laboratory has a number of KheperaII robots, LADAR sensors, and vision sensors.

    Specialties
    Robotics
    Sensing and sensor systems
    Design
    Ethics in Engineering
    neural networks
    Manufacturing

    TEACHING (Fall 2009)

    ME 270.01, ROBOT CONTROL/AUTOMATION, TuTh 02:50 PM-04:05 PM

    TEACHING (Spring 2010)

    ME 125L.001, CONTROL SYSTEMS,
    ME 125L.01L, CONTROL SYSTEMS,
    ME 125L.02L, CONTROL SYSTEMS,
    ME 125L.03L, CONTROL SYSTEMS,
    ME 125L.04L, CONTROL SYSTEMS,

    Recent Publications More Publications

    1. Kumar, M., Garg, D., and Zachery, R., A Method for Judicious Fusion of Inconsistent Multiple Sensor Data, IEEE Sensors Journal, vol. 7 no. 5 (2007), ppt. 723-733
    2. Kumar, M., Garg, D., and Zachery, R., Multiple Mobile Agent Control via Artificial Potential Functions and Random Motion, , (2007)
    3. Kumar, Manish and Garg, Devendra and Zachery, Randy, Stochastic adaptive sensor modeling and data fusion, Proceedings of SPIE - The International Society for Optical Engineering, vol. 6174 I (2006), ppt. 61740 - , [12.658478] [abs]
    4. Kumar, M. and Garg, D.P. and Zachery, R.A., A generalized approach for inconsistency detection in data fusion from multiple sensors, 2006 American Control Conference (IEEE Cat. No. 06CH37776C), (2006), ppt. 6 pp. - [abs]
    5. Kumar, M. and Garg, D.P. and Zachery, R., Multi-sensor fusion strategy to obtain 3-D occupancy profile, IECON 2005. Thirty-First Annual Conference of the IEEE Industrial Electronics Society (IEEE Cat. No.05CH37699), (2005), ppt. 6 pp. - [abs]

    Research Interests

      The main emphasis of research in the Robotics and Manufacturing Automation Laboratory is on the control of multiple robots that can work together. Multiple robotic control of the two ABB industrial robotic arms was sponsored by the National Science Foundation. The sensor modeling, data acquisition, data management, and sensor fusion research was being funded by the Army Research Office and the National Science Foundation. We are currently exploring ways of using neural networks and fuzzy logic algorithms incorporated in the robot control strategies. In addition, we are emphasizing swarm intelligence and control in our research using a network of small-size mobile robots. The research is primarily inspired by the existence of very robust biological counterparts such as swarming in ants, flocking of birds, and schooling in fishes. The primary feature of these systems that has attracted researchers is that the intelligence associated with an individual agent (e.g., ant or bird) is very primitive, and it utilizes interactions at local level to arrive at very simple decisions. This behavior at local level emerges into a group behavior that appears to be very robust and complex. This observation in biological systems has led the interested scientists and engineers to investigate multiple agent cooperative controls problem using a bottom-up approach. We have developed a multiple mobile robot test bed equipped with a Cognex-5400 camera, 2 SICK LADAR sensors, and 8 KheperaII mobile robots. These mobile robots are controlled via a radio controller through a desk-top computer. We are also working on developing formation control algorithms and strategies. The next step will be to implement these algorithms on our test-bed. The control of robotic devices via the internet has become an increasingly important area of research in the last few years. We have created a web based interface for our ABB IRB 140 industrial arms that provided the user with various functionalities, such as moving the robot arms linearly to specified coordinate offsets, opening and closing the grippers, rotating the tool (gripper) about the three axes (x, y and z), accessing the F/T values, and moving the conveyor and the indexing table. The major emphasis of our research on work cell simulation is on machine tools and related hardware operating in flexible manufacturing work cells. Past problems and recent advances, and guidelines for work cell design were also looked at. Two flexible manufacturing work cell models were created, which are capable of manufacturing a certain part. The costs of each of the layouts were compared with the costs of manufacturing the part to determine the optimal layout. Our research work in sensor fusion involves development of formal approaches to capture uncertainties inherent in sensor measurements in the form of appropriate probabilistic and analytical sensor models, and use those models to fuse data from multiple sources. The uncertainties involved in sensor measurements can arise from each sensor’s limitations, change in environmental parameters, or performance of estimation/calibration algorithm (such as image processing algorithm in case of vision sensor). The research focuses on developing a unified approach to capture uncertainties arising from any possible source in the form of sensor models, and involves the use of multiple vision sensors, infra-red sensors, and sonar ranging sensors .

    The mission of Duke's Mechanical Engineering and Materials Science educational programs is to provide the knowledge, skills, and credentials needed to be successful in the practice of engineering; the preparation necessary to undertake professional registration; an educational preparation for graduate or professional study; and an education background that is the basis for professional growth and leadership throughout a career that may encompass a broad range of endeavors, both technical and non-technical.