Robotics Research - GRASP ModLab  UPenn

This project develops a 2.5D simulation and control framework for COMPA/HAMR, a holonomic off-road robot with a differential drive base and 3-DoF gimbal turret. The system integrates an MPC controller with holonomic and traversability constraints. It uses a local traversability opttimizer and ARA* for global planning, Jacobian-based control, and slope-aware traversability layers for stable motion over uneven terrain. The work includes ROS 2 (Jazzy) advanced integration in C++ and Python, terrain mapping with Grid Map and Gazebo heightmaps, and hardware validation on a rocker-suspension with a mounted stabilizing gimbal prototype.

Submitted to ICRA 2026 in collaboration with George Mason University (GMU).

Future work: extend to a Holonomic Traversability Optimizer for optimal autonomous motion in 2.5D rough terrains, improving path smoothness, energy efficiency, and terrain adaptivity. GitHub Repos:

• High level Controller and Simulation: https://github.com/cedrichld/hamr_holonomic_robot

• Low level Controller (Embedded): https://github.com/virmani11kartik/HAMR_Controller

MiniP is a micro-aerial platform developed at ModLab that holds a Guinness World Record as the world’s smallest self-powered drone. I contributed to the aerodynamic performance effort, focusing on modeling, testing, and optimization of micro-propeller geometry. My work supported improved lift-to-drag characteristics leveraging Bayesian Optimization to drive propeller geometry improvements which maximized L/D at small scales. This helped push MiniP toward more stable, higher-efficiency flight within severe size, mass, and power constraints.