Job Number: P25INT-69
Honda Research Institute USA (HRI-US) is seeking a self-motivated research intern to advance policy learning and action grounding for multimodal embodied agents. The intern will investigate methods that enable intelligent agents to connect multimodal perception, task context, and prior knowledge with effective actions in physical or simulated environments. The research may encompass policy and skill acquisition, multimodal representation learning, learning from demonstrations or interaction, and generalization across tasks, environments, and embodiments. The specific research questions will be shaped jointly based on evolving project priorities and the candidate’s expertise, with the goal of producing high-quality research and advancing the capabilities of embodied agents.
San Jose, CA
- Formulate and investigate research problems related to policy learning, action grounding, and multimodal embodied intelligence.
- Develop learning methods that connect multimodal observations and task information to goal-directed, executable behavior.
- Explore different sources of experience and supervision, potentially including human demonstrations, robot data, simulated interaction, language, and large-scale multimodal datasets.
- Design and evaluate models for acquiring reusable policies, skills, or action representations that generalize across tasks and environments.
- Build experimental pipelines and conduct systematic evaluations in simulation and/or on robotic platforms.
- Analyze experimental results, identify research insights, and communicate findings through presentations, technical reports, demonstrations, research publications, and patents.
- Collaborate with researchers across machine learning, computer vision, multimodal foundation models, and robotics.
Minimum Qualifications
- Currently enrolled in a Ph.D. program in Computer Science, Robotics, Electrical or Computer Engineering, Machine Learning, or a closely related field.
- Strong foundation in machine learning, deep learning, computer vision, robotics, or multimodal learning.
- Proficiency in Python and experience with a deep-learning framework such as PyTorch or JAX.
- Research or substantial project experience in at least one relevant area, such as robot learning, embodied AI, multimodal learning, video understanding, imitation learning, or reinforcement learning.
- Ability to formulate research questions, implement and evaluate learning systems, and analyze experimental results.
- Strong written and verbal communication skills and the ability to work effectively in a collaborative research environment.
Bonus Qualifications
- Proficiency with one or more of the following: robot learning, learning from demonstrations, imitation learning, reinforcement learning, multimodal foundation models, vision-language-action models, or world models.
- Familiarity with policy representations, skill learning, action representations, hierarchical decision-making, or grounded multimodal reasoning.
- Experience working with video, egocentric observations, human behavior data, robot demonstrations, simulation data, or other forms of embodied experience.
- Familiarity with robot manipulation, human-to-robot knowledge transfer, cross-embodiment learning, or generalization across tasks and environments.
- Experience with robotics or simulation platforms such as Isaac Sim, MuJoCo, ManiSkill, RLBench, PyBullet, ROS, or comparable systems.
- A record of research publications at relevant venues like CVPR, NeurIPS, ICLR, ICRA, ECCV, ICCV, CoRL etc, open-source implementations, and substantial research projects in a relevant area.
Years of Work Experience Required
0
Desired Start Date
1/11/2027
Internship Duration
3 Months
Position Keywords
Robot Learning, Embodied AI, Multimodal Foundation Models
