The Indian Institute of Technology Bombay has developed Adaptive Modality-guided Visual Grounding (AMVG), a model that enables machines to understand satellite images using natural language prompts.