Allan Jabri

Papers

Total Citations

176

H-Index

About

Allan Jabri is a researcher at the forefront of artificial intelligence and robotics, specializing in visuomotor control, reinforcement learning, and multi-object manipulation. His most influential work introduces **Universal Planning Networks (UPN)** (2018, 92 citations), a groundbreaking framework that embeds differentiable planning directly into goal-directed policies. This innovation enables agents to learn abstract representations for complex tasks, bridging the gap between planning and generalization in visuomotor control—a critical step toward more adaptable AI systems. Jabri also made significant strides in **relational reinforcement learning** for robotics, as demonstrated in his 2020 paper (63 citations) and its 2019 precursor (21 citations). These works address the notorious challenge of sparse rewards in multi-object manipulation, proposing scalable methods that drastically reduce data requirements. By focusing on relational reasoning, Jabri’s research empowers robots to handle increasing object complexity with practical efficiency. His contributions have been widely recognized, with cumulative citations exceeding 176, underscoring their impact on both theoretical foundations and real-world robotic applications. For students and researchers, Jabri’s work exemplifies how integrating planning, learning, and relational structures can unlock more intelligent and autonomous systems.