Welcome to my Webpage
I am a fourth year Ph.D. student at the Department of Mechanical Engineering, Texas A&M University, working under the supervision of Dr. Srikanth Saripalli at the Unmanned Systems Lab. My research interests are focused on Mobile Robot Localization and Mapping. I have a MS in Control Systems Engineering from École Centrale de Nantes, France and I did my MS Thesis at LAAS-CNRS, Toulouse, under the guidance of Dr. Antonio Franchi on Aerial+Ground Co-Manipulation. I have a B.Tech. in Electrical Engineering from National Institute of Technology-Rourkela, India.
Contact
Email: subodh514@tamu.edu
Office: ENPH 221, Texas A&M University-College Station
Current Research
- SURVIVABLE NETWORKS OF ROBOTS FOR ADVERSITY: RECOVERABILITY AND EXPENDABILITY: This project deals with usage of cheap and expendable robots for exploration and information gathering in an adversarial environment and how the probable death of the autonomous agent can help to formulate better planning algorithms for those robots hat follow the first one. We formulate this problem as a MDP(Markov Decision Process). This projects examines a model of a task where robot mortality is presupposed, and examines how destruction of a robot serves as an information source. The basic setting is that a robot operator has a swarm of cheap disposable robots; these are deployed in sequence, each performing a hazardous navigation mission, either completing their mission or failing to return. Situational awareness is improved in in either case: the operator uses robot failure, including destruction, serves as an information source. In this video one can see the drone flying over an adversary and returning to a safe zone after collecting information. The fact that it has returned safe is also a source of information to someone who is writing the planning algorithm.
- Augmented SLAM with Multi-Sensor Online Calibration In this project we aim to add the extrinsic calibration parameters of all the sensors to the SLAM(Simultaneous Localization and Mapping) state vector because in almost every real life scenario the calibration parameters change over time due to environmental effects on the autonomous agent. Therefore, adding the calibration parameters to the state vector will not only improve the calibration results but also improve the the result of localization and mapping.
Past Experience
- 2D-SLAM using RGB-D sensor: This work dealt with using a cheap RGB-D sensor on a differential robot to do 2D Occupancy Grid based Graph SLAM in indoor environment. I was involved this project as an intern during summer 2018 with the Robot Software Development team at Perceptin(now called Trifo) .
- Cooperative Aerial-Ground Manipulation: I worked on this project for my MS Thesis at LAAS-CNRS, Toulouse, under the guidance of Dr. Antonio Franchi. This work dealt with using a Ground Manipulator and a novel Tilted Rotor Aerial Manipulator to lift long bars in warehouses. The project had numerous sub-systems and my job was to integrate the various software and hardware components and also work on vision based localization techniques which remove the system's dependence on Motion Capture System. A demo of the entire system operation can be found here.