Kalesha Bullard

Georgia Institute of Technology, Meta (Israel)

Papers

Total Citations

H-Index

About

Kalesha Bullard is a roboticist whose research lies at the intersection of human-robot interaction (HRI), interactive learning, and multi-agent communication. Her work focuses on enabling robots to autonomously adapt to new environments and tasks through natural, human-guided interactions. A central contribution is her investigation of how robots can learn from demonstration, not just by mimicking actions, but by grounding task parameters in the physical world—a problem she tackled in her 2016 paper (12 citations). She also pioneered a human-driven approach to feature selection for learning classification tasks, allowing non-expert users to guide a robot's learning process (2018, 12 citations). To make these interactions more natural, Bullard developed a multimodal, real-time contingency detection system (2014, 12 citations) that helps robots recognize when a person is responsive to an interaction request, inspired by human cognitive processes. Most recently, her 2021 work on quasi-equivalence discovery for zero-shot emergent communication (6 citations) pushes into multi-agent settings, exploring how agents can develop shared communication protocols without prior training. Across her publications, Bullard consistently addresses the core challenge of deploying general-purpose robots that can learn and communicate effectively in the real world.