VLM-based Scene Understanding Research – Intern

Build up an advanced cloud-based system for sparse semantic scene understanding using VLMs.Architect a cloud-retrieval pipeline consisting of scene representation storage, localization and data retrieval, as well VLM-based map creation.Implement the system using off-the-shelf building blocks for localization, mapping and communication, extending them where needed.Summarize research findings in high-quality paper and/or patent submissions.