🛠️ Projects Portfolio
Here’s a showcase of selected research projects across robotics, AI, and healthcare — blending real-world application with academic rigor.
🧭 Autonomous Navigation with Deep Reinforcement Learning
UdG Girona, Spain | Feb 2024 – Jun 2024
An end-to-end learning pipeline for autonomous navigation using Soft Actor-Critic (SAC) and PPO, built with a custom Gymnasium environment and LIDAR-based input.
🎥 Watch Project Video
📄 Read Thesis PDF
🤖 Autonomous Robotic Explorer & Manipulator
UdG Girona, Spain | Feb 2023 – Jul 2023
An integrated robotic system combining exploration, manipulation, SLAM, and perception, developed in ROS, tested in simulators and real environments.
📸 Project Snapshots
🔹 Planning
- Frontier-based exploration combining RRT and DWA
- ✅ Tested in Gazebo and Stonefish
📄 Read PDF
🔹 Localization
- Pose-based EKF SLAM with ICP scan matching
- 🔄 Sensor fusion using LiDAR and IMU
📄 Read PDF
🔹 Perception
- Event-based tracking for DAVIS sensor
- ✅ Integrated RealSense + ArUco detection
🎥 Watch Video
📄 Read PDF
🔹 Manipulation
- Task-priority control for 4-DOF robotic arm
- 🤖 Autonomous pick-and-place
📄 Read PDF
🔹 Autonomy
- Behavior Trees using
py_trees - 🧠 Seamless decision-making
🎥 Real Robot
🎥 Simulation
🧬 Medical Imaging Quality Assessment using Radiomics
University of Barcelona | Jul 2023 – Aug 2023
Evaluated synthetic imaging realism using Radiomics FID. Analyzed robustness to noise, correlation with clinical features, and limitations of classical image realism metrics.
🌐 Pomona 3D Graph SLAM
Budapest, Hungary | Sep 2023 – Dec 2023
Developed a 3D mapping solution in ROS using LiDAR, IMU, and GPS, fusing data into optimized pose graphs using hdl_graph_slam. Focused on accuracy, real-time loop closure, and outdoor readiness.
💬 Multi-Span Medical QA (QueSemKnow)
IIT Patna, India | Jun 2022 – Aug 2022
Developed QueSemKnow, a query semantic and knowledge-guided transformer model for multi-span question answering in medical settings. Published at LREC-COLING 2024.