Apart from my professional experiences, I have taken part in several competitions and explored other personal projects. This page will give you a brief overview of all my projects and the associated documentation.
- Target-pose Aware Grasping Framework
- Developed Neural Descriptor Field (NDF)-based SE(3) equivariant grasping framework conditioned on the desired target pose with a Franka Panda arm in pybullet simulation.
- Enabled pick and place of objects from arbitrary initial and final poses without demonstrations with a grasp success rate of 72.34%
- Synergistic pushing and grasping framework using self-supervised deep reinforcement learning
- Jointly trained two Densenet-121-based CNNs in a self-supervised Q-learning framework in simulation.
- Developed scaffolding mechanism to facilitate weight fine-tuning with with a Franka Panda arm
- Achieved grasp success rate 73% and push-then-grasp success rate of 65%
- Autonomous Object Rearrangement Robotic Arm (Nov 2020 – June 2021)
- Integrated perception, grasping, motion planning and control sub-systems on an autonomous object rearrangement robot.
- Implemented a primitive failure tolerance framework to handle common edge-cases in table organization problem.
- MedBot: Sponsored Project (Jun 2020-Aug 2020)
- Led system design and integration of software, electronics and mechanical systems of a WiFi-controlled food-delivery & UV-sanitation service Robot for KIMS Hopitals’ COVID-19 ward.
- Executed unit, integration and acceptance testing prior to deployment in hospital ward.
- Hydraulically Actuated Nurse (HyAN) : Undergraduate Capstone Project (Jul 2019-Apr 2020)
- Led mechanical design, fabrication and static analysis modules for lifting mechanism to carry payloads up to 175 kg.
- Deployed Inception-v3 based deep learning model to classify safe and unsafe patient lifting poses in real-time with 97% accuracy.
- Finished top 0.5% at Smart India Hackathon with 250,000+ students, 3000+ Institutions & 200+ organizations.
- Biomimetic Robotic Fish: e-Yantra Robotics Competition (2019-20 Oct 2019-Mar 2020)
- Headed the design of actuation mechanism, enclosures, and PCB for robotic fish and RF controller on ground station.
- Implemented servo motor control loop to facilitate flapping of caudal fins to achieve carangiform locomotion.
- Finished 4th Place (top 1%) at India’s largest Robotics Competition with 30000+ participants.
- Hungry Bird: e-Yantra Robotics Competition 2018-19 (Oct 2018-Mar 2019)
- Implemented a PID & LQR controller to control altitude, roll, pitch and yaw of a drone to facilitate take-off and hovering at a given set-point.
- Implemented elementary state-machine and RRT-Connect algorithm using Open Motion Planning Library to emulate feeding patterns of birds in simulation (CoppeliaSim) and real-world with a Pluto-X drone.
- Spherical Drone (Dec 2018-Mar 2019)
- Built experimental drone to make it easier to fly through narrow spaces and roll across uneven terrain.