Skip to content
Mihir Kulkarni
  • Home
  • Publications
  • Experience
  • Talks
  • Gallery
  • Projects
    • Englishहिंदीमराठी

Mihir Kulkarni

I am a Postdoctoral Researcher at the Norwegian University of Science and Technology (NTNU), where I completed my Ph.D. in Engineering Cybernetics in 2025 under the supervision of Prof. Dr. Kostas Alexis and co-supervision of Prof. Dr. Davide Scaramuzza. My PhD focused on vision-driven navigation and control for multirotors in complex environments, massively parallelized simulation and rendering and sim2real transfer techniques.

I developed the Aerial Gym Simulator, a massively parallelized simulation and rendering framework for multirotors (now integrated into NVIDIA Isaac Lab), and co-developed UniPilot and the Unified Autonomy Stack, a modular hardware-software autonomy stack for diverse robots.

I was a key member of the winning Team CEREBRUS in the DARPA Subterranean Challenge. I led the artifact detection subsystem and co-developed a multi-robot exploration path planner for heterogeneous robot teams. This endeavor culminated in a USD 2mn prize and involved integrating and deploying complex multi-robot teams for robust search-and-rescue operations in harsh, unstructured underground environments.

Education

Ph.D. in Engineering Cybernetics, NTNU, Trondheim, Norway (2025)
M.S. in Computer Science and Engineering, University of Nevada, Reno, USA (2021)
B.E. in Mechanical Engineering, BITS Pilani, Goa, India (2020)

Thesis | Google Scholar | Download CV

Mihir Kulkarni
Postdoctoral Researcher
NTNU

github google_scholar linkedin X

Featured publications

img/projects/aerial-gym.jpg

Aerial Gym Simulator

A framework for highly parallelized simulation of aerial robots based on NVIDIA Isaac Gym.
img/projects/uas.jpg

Unified Autonomy Stack

A field-tested autonomy architecture commanding a diverse set of robots.
img/projects/collision-encoding.jpg

Reinforcement Learning for Collision-free Flight

Exploiting Deep Collision Encoding for autonomous flight in dense environments.
img/projects/se-dcp.jpg

Semantically-enhanced Deep Collision Prediction

Autonomous navigation using aerial robots with semantic awareness for collision prediction.
img/projects/cerberus.jpg

DARPA Subterranean Challenge

Winning team (Team CERBERUS) of the DARPA Subterranean Challenge.
img/projects/unipilot.jpg

UniPilot: Enabling GPS-Denied Autonomy Across Embodiments

A unified pilot for enabling GPS-denied autonomy across different robotic embodiments.
All projects ⟶
Copied! Copy code to clipboard
  • feed
  • email
  • github
  • google_scholar
  • linkedin
  • X
  • sitemap
Powered by Zola & tabi • Site source

Search

No results $NUMBER result $NUMBER results $NUMBER results $NUMBER results