Srinivas Devadas

Massachusetts Institute of Technology, Artificial Intelligence in Medicine (Canada)

Papers

Total Citations

H-Index

About

Srinivas Devadas is a pioneering researcher at the intersection of computer architecture and robotics, where his work redefines how domain-specific hardware can meet the extreme computational demands of autonomous systems. His key research areas include robomorphic computing, hardware-software co-design, and parallel acceleration for robotics. Devadas introduced the concept of **robomorphic computing**, a design methodology that tailors domain-specific accelerators to the unique morphology and real-time constraints of robots, addressing a critical performance gap in motion planning and control. His highly cited work on accelerating robot dynamics gradients across CPUs, GPUs, and FPGAs (39 citations) demonstrates how parallel platforms can unlock order-of-magnitude performance gains for state-of-the-art control algorithms. Through rigorous benchmarking and workload analysis of rigid body dynamics (16 citations), he has provided foundational insights for emerging nonlinear control techniques. With over 42 citations on his robomorphic computing framework alone, Devadas’s contributions are shaping the future of efficient, hardware-aware robotics, bridging the gap between theoretical algorithms and practical, real-time deployment in complex physical systems.