Research Scientist Internship-Physical AI 개발

The internship focuses on developing robot understanding of visual information and language instructions to enable real-world actions. Responsibilities include building large-scale robotics training data, simulation-based training and validation, evaluating in real robot environments, improving generalization across environments and tasks, and researching novel architectures and learning methods (SFT, RL, data mixing, data filtering). It also covers automating end-to-end evaluation, Sim-to-Real verification, and leveraging diverse data sources (teleoperation, demonstrations, simulation) to improve model performance. The role involves designing in-house benchmarks and task suites, collaborating with LG AI researchers, and contributing to the Robot Foundation Model, Vision-Language-Action (VLA), World Model, and Embodied AI initiatives.