remote
Research Scientist - Robotics Foundation Models - Honda Research Institute USA
Research Engineer
Lead research on multimodal foundation models for dexterous robotic manipulation, integrating vision‑language understanding with learning‑based control to create reliable, adaptable robotic systems.
About the role
Key Responsibilities
- Design, train, and fine‑tune multimodal foundation models (VLMs, VLAs) tailored for robotic manipulation tasks.
- Develop vision‑language perception pipelines that enable robots to interpret and act on complex instructions.
- Integrate learned models with model‑based control to achieve robust, failure‑aware dexterous manipulation in real‑world settings.
- Collaborate with cross‑disciplinary teams to evaluate system performance and iterate on architecture improvements.
- Publish findings in top robotics and AI venues, contributing to the broader research community.
Requirements
- Ph.D. or equivalent experience in robotics, computer vision, or machine learning.
- Proven expertise in foundation models, vision‑language architectures, and deep learning frameworks.
- Strong programming skills in Python and experience with PyTorch or TensorFlow.
- Experience with robotic simulation and real‑world deployment of manipulation systems.
- Excellent communication skills and a track record of publishing high‑impact research.
Skills
deep learningreinforcement learning