David Matthews

Northwestern University, University of Vermont

Papers

Total Citations

H-Index

About

David Matthews is a robotics and artificial intelligence researcher whose work sits at the intersection of automated robot design, embodied cognition, and natural language understanding. His most influential contribution, "Efficient Automatic Design of Robots" (2023, 27 citations), tackles one of robotics' most persistent challenges: the manual, time-intensive process of designing robots with complex interdependencies between structure, sensing, and behavior. By advancing automated design pipelines, Matthews is helping to democratize and accelerate robot development. His follow-up work, "Evolution and Learning in Differentiable Robots" (2024), pushes this frontier further by leveraging massively parallel differentiable simulation to overcome longstanding bottlenecks like premature convergence and poor sim-to-real transfer. Equally notable is Matthews' pioneering research on language grounding in embodied systems. His 2019 paper "Word2vec to Behavior" explored how word embeddings can help robots interpret and act on natural language commands, a thread continued in his work on crowd grounding through human-robot interaction. Together, these contributions position Matthews as an emerging voice in autonomous robot design and machine language understanding, with growing influence across the robotics research community.