Eadom Dessalene

George Mason University, University of Maryland, College Park

Papers

Total Citations

H-Index

About

Eadom Dessalene’s research lies at the intersection of robotic manipulation, computer vision, and language-guided control, with a focus on enabling robots to grasp objects and follow instructions with greater precision and adaptability. In his highly cited 2019 work, Dessalene introduced novel geometric feature representations that capture hand-object relationships during the critical near-contact stage of grasping—before the fingers close. These features proved robust to noise in joint and pose variation, offering a stable foundation for more reliable robotic grasping. His 2020 paper advanced instruction-following in reinforcement learning by proposing a method that learns to imagine and reach visual goals, bypassing the need for extensive prior linguistic or perceptual knowledge. This end-to-end approach allowed policies to map observations and instructions directly to actions, marking a significant step toward more intuitive human-robot interaction. With 13 citations across his top papers, Dessalene’s work is shaping how robots perceive and act in unstructured environments, bridging geometric reasoning and goal-directed behavior.