Research Engineer (Vision Language Navigation/Exploration)

Job Description

This project focuses on long-horizon embodied navigation for heterogeneous multi-robot collaboration in previously unseen, partially observed, and open-ended environments, where different types of embodied agents must efficiently locate target by sharing observations, spatial memory, and navigation plans. Guided by multimodal goal inputs, agents are expected to ground goals, perceive their surroundings, leverage spatial memory, reason about spatial relationships, and identify the correct target during navigation. Throughout this framework, we assume agents can continuously update their representation of the environment through online observations and collaborate with other agents that may have different sensing, mobility, or task capabilities. These tasks will be approached using vision-language models and multimodal foundation models, enabling agents to perform open-vocabulary perception, multimodal goal grounding, spatial reasoning, and high-level planning within a deployable embodied navigation framework.

Qualifications

• Possess a master’s degree in Computer Engineering, Robotics, or a closely related discipline.
• Strong coding skills in Python and experience with deep learning frameworks such as PyTorch.
• Experience with simulation environments or benchmarks such as Habitat, Isaac Sim, GOAT-Bench, etc.
• Publications or research experience in VLN, robotics, VLMs/LLMs, reinforcement learning, or related areas.
• Literature review/summarizing skills
• Good writing/spoken communication skills