Robotics, SLAM, Localization, AI, Reinforcement Learning
Resume:
Bharath Masetty
*** ** * ******, ********* ***, Austin, TX
Q adjwsn@r.postjobfree.com bmasetty BharathMasetty Portfolio Education
+ The University of Texas at Austin
M.S.(Thesis) in Mechanical Engineering GPA 3.89/4.00 Focus Areas: Robotics, Reinforcement Learning & Controls 2019-2021
+ Indian Institute of Technology (IIT) Kharagpur GPA 8.72/10.0 B.Tech. (Hons.), Department of Mechanical Engineering 2015-2019 Projects
Reinforcement Learning based Dynamic Stabilization for NAO Humanoid Robot C++ Python
+ Implementing Whole Body Operational Space Control (WBOSC) based kick on PyBullet simulator as a baseline.
+ Working on an RL based control policy for single-leg balancing in kick and walk motions on Nao bipedal robots.
+ Devising on sim-to-real approach to transfer the policy on the simulator to the real robot with limited compute power. Reinforcement Learning for Human Motor Skill Acquisition Python
+ Developed ReachNinja, a gym-like learning environment for reinforcement learning agents and human motor learning.
+ Developing an RL based curriculum generating agent to create an adaptive motor skill training framework for humans.
+ Conducting human experiments using handcrafted curricula to compare against direct target training as a baseline. Autonomous Mobile Robotics [Report] C++
+ Successfully replicated the implementation of the dynamic pose graph slam algorithm on a 1/10th scale mobile robot.
+ Developed a global navigation stack with A* path search, lattice based navigation graph, and local obstacle avoidance.
+ Implemented a particle filter algorithm for localization and mapping with correlative scan matching using GTSAM. Action Modeling using Modular Inverse Reinforcement Learning [Report] Python
+ Retrieved the reward function from human game-play actions on Atari games from Atari-HEAD dataset.
+ Formulated a modular decomposition of human reward structure for sample efficient policy optimization.
+ Reproduced the game play behaviour using the resultant reward function on the arcade learning environment (ALE). Characterizing Narrative Comprehension along Principal Gradients of the Brain [Report] Python
+ Evaluated the functional connectivity gradients using diffusion embedding on fMRI data for multiple auditory stimuli.
+ Analysed the changes in the functional hierarchy of different cortical regions for intact and scrambled narratives.
+ Presented a novel interpretation of fMRI data in the presence of real-life stimuli using the principal gradients. Human Motor Learning as a Reinforcement Learning Process: A Review [Report]
+ Modeled human motor learning process as a combination of model-based and model-free reinforcement learning.
+ Analysed temporal resolution of cerebellar credit assignment using a simulated cerebellum on cartpole control task. Publications
Conferences
+ Keya Ghonasgi, Reuth Mirsky, Sanmit Narvekar, Bharath Masetty, Adrian Haith, Peter Stone, and Ashish Deshpande. Capturing Skill State in Curriculum Learning for Human Skill Acquisition. International Conference on Robotics and Automation 2021. [under review]
+ Bharath Masetty, Reuth Mirsky, Ashish Deshpande, Michael Mauk, and Peter Stone. Is the Cerebellum a Model-Based Reinforcement Learning Agent?. In Proc. of the 20th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2021), London, UK, May 3–7, 2021,IFAAMAS, 5 pages. [under review] Referred Workshop & Symposia
+ Keya Ghonasgi, Reuth Mirsky, Bharath Masetty, Sanmit Narvekar, Adrian Haith, Peter Stone, and Ashish Deshpande. Leveraging Reinforcement Learning for Human Motor Skill Acquisition. In Social AI for Human-Robot Interactions of Human-Care Service Robots Workshop at the International Conference on Intelligent Robots and Systems (IROS), Las Vegas, Nevada, October 25-29, 2020.[Paper]
+ Suna Guo, Bharath Masetty,Ruohan Zhang, Dana Ballard, Mary Hayhoe. Modeling human multitasking behavior in video games through modular reinforcement learning.Vision Science Society 2020. [Paper] Relevant Courses & Skills
Courses Machine Learning Reinforcement Learning Brain-Body & Robotics Neural Computation Autonomous Robots Deep Learning Humanoid Robot Control Neural Control Languages: C++, Python Platforms: Linux, Windows
Robotics & AI Platforms: ROS, PyBullet, Open AI Gym, Gazebo, ALE, pytorch, MATLAB, SolidWorks
Contact this candidate