Affordance Learning

Affordance learning focuses on enabling artificial agents, particularly robots, to understand and utilize the potential actions (affordances) offered by objects and environments. Current research emphasizes grounding affordances in 3D space using vision-language models (VLMs), often incorporating techniques like potential fields, Gaussian splatting, and attention mechanisms, to predict interaction regions and generate appropriate actions from natural language instructions. This field is crucial for advancing robotics, particularly in areas like human-robot collaboration and autonomous navigation, by enabling robots to interact more intelligently and flexibly with their surroundings.