The work below was done with these groups. Space robotics is not a solo discipline: hardware that has to survive a field campaign takes a lab, and the interesting problems sit where two groups' expertise meet.
Genova, Italy
With Claudio Semini
Learning-based locomotion controllers for quadrupedal robots in space exploration, under the ALTEC project. GPU-accelerated reinforcement learning and sim-to-real transfer for deployment on legged hardware.
Sendai, Japan
With Kazuya Yoshida, Kentaro Uno, Shreya Santra, Elian Neppel
MoonBot, a modular reconfigurable robot for lunar base construction, developed under JST Moonshot Goal 3. Doctoral research on decentralized control and reinforcement learning for robots whose morphology changes.
Sagamihara, Japan
Three field campaigns in lunar-analogue terrain and lighting: manipulator deployment for solar infrastructure, reinforcement learning policies on morphology-varying robots, and full-scale validation of the MoonBot APEX planning framework.
MoonBot APEX: Agentic Planning and Execution FrameworkModular RL Policy Implementation on Reconfigurable RobotsEnhanced Manipulator Control for Solar Infrastructure Deployment
NASA Johnson Space Center and Rice University, Houston, USA
With Jacob Cohen, Chief Scientist at NASA Ames, as supervisor
Led the project management team and served as lead editor for the International Lunar University project during the Space Studies Program 2024, hosted at NASA Johnson Space Center and Rice University. Later presented at the International Astronautical Congress 2024.
India
An AI-enabled satellite-based air pollution monitoring and alert system, developed with NRSC-ISRO scientists during the Space Studies Program. Recognised with the I-CON Award from the Indian Society of Remote Sensing.
New Delhi, India
A central cloud platform for controlling and managing a cluster of electric vehicle charging stations, built with EVI Technologies.
Delhi, India
With Manoj Saxena, Geetika Jain Saxena
Semiconductor device modelling and secure signal processing: machine-learning models for GaN HEMT parameter estimation and radiation damage, and memristor-based approaches to watermarking and cryptographic processing.