Debarshi Patanjali Ghoshal

Past Research Experience

  • M.Tech thesis at Department of Electrical Engineering, Indian Institute of Technology Kanpur:
    Robot Learning from a Human Expert through Modified Kinesthetic Teaching
    SUPERVISOR: Dr. Laxmidhar Behera
    (Professor, Department of Electrical Engineering, IITK.)
    Description: Teaching new motor tasks to robots through physical interactions is an important goal for both robotics and machine learning. Most monolithic machine learning approaches fail to scale when going beyond basic skills. In this work we present a simple framework for teaching the robot (to hit the ball) through direct physical interaction with a human teacher (i.e. Kinesthetic Teaching). Current popular established method of kinesthetic teaching generally uses a two-stage approach: First, a library of motor primitives is generated through direct physical manipulation of the robot. In second stage, a reinforced learning ("reward" stage) is implemented to dynamically adjust the policy of choosing from motor primitive library. In this work, we modify the Kinesthetic Teaching by incorporating the domain experience of the human teacher through repeated teachings, instead of relying on the "reward" stage. This approach has advantage: One potential problem with the "reward" learning phase is that there may be subjective difference of what is a "good" shot from the kinesthetic teaching and from the bystander viewpoint (later in "reward" stage). Even a little difference in this regard will result in a confusing feedback ("reward"), and hence it would be difficult to correctly figure out which library training samples should be reassigned what weight values. Our approach eliminates this problem altogether.
    The work resulted in a peer-reviewed international conference paper.

  • Summer Internship at Indian Institute of Science, Bangalore:
    Waypoint Navigation System for Unmanned Aerial Vehicles (UAV)
    SUPERVISOR: Dr. Seetharama M. Bhat
    (Chairman, Department of Aerospace Engineering, IISc.)
    Description: I worked on Waypoint navigation system of Unmanned Aerial Vehicles (UAV). I acquainted myself with the aerodynamics of UAVs, and designed a controller consisting of navigation, guidance, and control loops about a dynamic model of a UAV. For evaluation, a Matlab Simulink model was built which yielded fairly good result. I was also involved in setting up the Hardware-in-the-Loop simulation system of Kestrel Autopilot System. The simulation was implemented successfully.
    IISc Project Report

  • Research Project:
    Neural Network Approach for Automatic Number Plate Recognition (ANPR)
    SUPERVISOR: Dr. Anjan Rakshit
    (Professor, Department of Electrical Engineering, Jadavpur University.)
    Description: Automatic Number Plate Recognition (ANPR) is a real time embedded system which identifies the characters directly from the image of the license plate. ANPR systems are very useful to the law enforcement agencies as the need for Radio Frequency Identification tags and similar equipments are minimized. Since number plate guidelines are not strictly practiced everywhere, it often becomes difficult to correctly identify the non-standard number plate characters. This problem was addressed by using a pixel based segmentation algorithm of the alphanumeric characters in the license plate. The non-adherence of the system to any particular country-specific standard & fonts effectively means that this system can be used in many different countries – a feature which can be especially useful for trans-border traffic e.g. use in country borders etc. The system was tested on 150 different number plates from various countries and an accuracy of 91.59% was reached.
    The work resulted in a peer-reviewed conference paper, which also won the best-paper prize of the conference.

  • Research Project:
    Robust Controller Design for Boiler Burning Process using RBode Plot
    SUPERVISOR: Dr. Smita Sadhu
    (Professor, Department of Electrical Engineering, Jadavpur University.)
    Description: Robust Bode (RBode) plot was used to design a robust controller for boiler burning process. The boiler plant was modeled as a first-order dynamic lag with time-delay. The designed controller was compared with a traditional Cohen-Coon controller. The significance of using RBode plot method in case of time-delay systems is twofold: (i) RBode method had not been applied on time-delay systems previously. Through this work, applicability of RBode technique with time-delay systems was established, which, in turn, widened the scope of application of this technique. (ii) Successful implementation of RBode method on time-delay systems indicated its potential in case of higher order systems (as many practical higher order systems may be approximated as a lower order system with time delay).
    The work resulted in a peer-reviewed conference paper.

  • Final-year Elective Project at Department of Electrical Engineering, Jadavpur University:
    Internal Model Control
    Description: Internal Model Control (IMC) is an interesting approach to Control Systems which differs from the traditional feedback control. Considering a typical first-order plant transfer function with time-delay, I designed a controller based on IMC scheme, and compared its performance with that of classic Ziegler–Nichols tuning based PID controller. I also investigated the robustness of the IMC controller by introducing real-world problems such as noise and process-model mismatch. Further performance evaluation of the IMC controller was done by considering higher-order plant transfer functions.
    Project Reports: 1 and 2


Back