Papers
7
Total Citations
72
H-Index
5
About
Radhika Ghosal is a leading researcher at the intersection of robotics, computer architecture, and safety-critical systems. Her work focuses on making autonomous robots—from micro aerial vehicles (UAVs) to ground-based platforms—more efficient, resilient, and deployable. Ghosal’s key contributions include **RoboShape**, a framework that uses topology patterns to scalably deploy hardware accelerators across diverse robots, addressing the critical challenge of non-recurring engineering costs. She also developed **GRiD**, a GPU-accelerated library for rigid body dynamics with analytical gradients, which accelerates trajectory optimization for robotic planning and control. Her most impactful work, **MAVFI** (over 22 citations), introduces an end-to-end fault analysis framework that assesses the effect of silent data corruption on UAV mission metrics like flight time and success rate, while incorporating anomaly detection and recovery. Ghosal’s **ROSFI** methodology further extends this resilience analysis to the Robot Operating System (ROS), using UAVs as a case study. Her research has been published at top venues including MICRO, ICRA, and IROS, and she is recognized for pioneering system-level approaches that bridge hardware acceleration, fault tolerance, and autonomous navigation.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3GRiD: GPU-Accelerated Rigid Body Dynamics with Analytical Gradients14 citations · 2022
- 4
- 5Rapid Prototyping of Pneumatically Actuated Inflatable Structures5 citations · 2019
- 6
- 7RoboRun: A Robot Runtime to Exploit Spatial Heterogeneity2 citations · 2021