David Harel

Papers

Total Citations

H-Index

About

David Harel is a distinguished computer scientist whose recent research sits at the intersection of deep reinforcement learning (DRL), formal verification, and robotics. His work addresses one of the most pressing challenges in modern AI: ensuring the safety and reliability of learning-based autonomous systems. Harel has made notable contributions to the verification of robotic navigation systems powered by deep neural networks, pioneering methods to detect and address bugs in DRL policies that govern complex reactive behaviors. His research on constrained reinforcement learning for robotics introduces scenario-based programming as a framework for enforcing safety constraints in high-stakes environments where human safety and costly hardware are at risk. By bridging the gap between DNN verification techniques and real-world robotic applications, Harel's work provides critical tools for deploying autonomous systems responsibly. His 2023 paper on verifying learning-based robotic navigation systems has already garnered 18 citations, reflecting the timeliness and influence of his contributions. For students and researchers working in safe AI, autonomous robotics, or formal methods, Harel's work offers foundational insights into making intelligent systems both capable and trustworthy.