Your application is being processed
Research Intern: Human-Object Interaction Learning
Job Number: P26INT-10
Honda Research Institute USA (HRI-US) is seeking a self-motivated and independent research intern to join our team researching multimodal machine learning for human-object interaction understanding. This role will focus on developing novel representation learning methods that capture how humans interact with and manipulate objects, leveraging egocentric video, hand pose, object-centric information, and language. The intern will investigate learning approaches that model hand-object relationships, manipulation dynamics, and temporal interactions to support downstream tasks such as action understanding, error detection, procedural activity understanding, and skill assessment. This research-oriented internship aims to develop novel methodologies and produce high-impact results suitable for publication at a top-tier conference.
San Jose, CA
|
Key Responsibilities
|
|
- Conduct literature reviews to identify relevant prior work, formulate research hypotheses, and propose novel technical directions in human-object interaction representation learning.
- Design and implement machine learning methods for learning representations of human-object interactions from egocentric video, hand pose, object-centric features, and language.
- Investigate multimodal representation learning approaches that capture spatial relationships, hand-object contact, manipulation dynamics, and temporal interactions.
- Explore methods for integrating human-object interaction representations into modern video understanding models and vision-language models.
- Develop and evaluate learned representations on downstream tasks such as action recognition, temporal action segmentation, error detection, procedural activity understanding, and skill assessment.
- Design rigorous experiments, evaluation protocols, and ablation studies to assess representation quality, generalization, and downstream task performance.
- Build data processing, model training, and evaluation pipelines for multimodal human-object interaction learning.
- Analyze experimental results, communicate research findings, and collaborate with researchers to develop novel methods suitable for publication at leading conferences such as CVPR, ICCV, ECCV, NeurIPS, and ICLR.
Minimum Qualifications
|
|
- Currently pursuing a Ph.D. in Computer Science, Electrical Engineering, Robotics, or a related field.
- Strong background in machine learning and deep learning, with hands-on experience in computer vision, representation learning, or multimodal learning.
- Proficiency in Python and deep learning frameworks such as PyTorch.
- Demonstrated understanding of one or more of the following areas: video understanding, human-object interaction, action recognition, human pose modeling, representation learning, or vision-language models.
- Experience designing and conducting empirical machine learning experiments, including data preprocessing, model training, evaluation, and ablation studies.
- Strong analytical and problem-solving skills, with the ability to understand, implement, and extend methods from recent research papers.
- Strong written and verbal communication skills, with an interest in conducting original research and publishing at leading academic conferences.
Bonus Qualifications
| Years of Work Experience Required |
0 |
| Desired Start Date |
May 2027 |
| Internship Duration |
3 Months |
| Position Keywords |
Human-Object Interaction, Egocentric Video Understanding,
Multimodal Learning, Representation Learning, Object Manipulation,
Vision-Language Models
|
|
|
|
Alternate Way to Apply
Send an e-mail to careers@honda-ri.com with the following:
- Subject line including the job number(s) you are applying for
- Recent CV
- A cover letter highlighting relevant background (Optional)
Please, do not contact our office to inquiry about your application status.