Internship - Foundation Models for Humanoid Robots

merl; Cambridge (MA); On-site; USD 67,000 - 89,000; Full time

Job summary

MERL is seeking a highly motivated Ph.D. student to research foundation models for humanoid loco-manipulation in factory automation, focusing on capabilities, safety, and robustness.

The internship will develop humanoid systems for material handling tasks in warehouse and manufacturing environments, collaborating with MERL researchers to integrate Vision-Language-Action models, world models, and robot control for robust long-horizon manipulation.

Qualifications

• Pursuing a Ph.D. in Mechanical Engineering, Computer Science, Electrical Engineering, or a related field.
• Hands-on research with Vision-Language-Action models and multimodal robotics.
• Experience with world models for planning, prediction, or decision making.
• Sim-to-real transfer and robotic simulation (Isaac Lab, MuJoCo).
• Strong publication record or potential in robotics/AI conferences/journals.

Responsibilities

• Research on foundation models for humanoid loco-manipulation in factory automation.
• Develop algorithms integrating VLA models, world models, and robot control.
• Collaborate with MERL researchers to publish at top conferences.
• Explore safety and robustness in long-horizon manipulation.

Skills

VLA models

Multimodal foundation models

World models

Sim-to-real transfer

Publications in top venues

Education

PhD candidate (Mechanical Eng / Computer Science / Electrical Eng)

Tools

Isaac Lab

MuJoCo

Job description

MERL is seeking a highly motivated Ph.D. student to conduct research on foundation models for humanoid robot loco-manipulation in factory automation, with a focus on understanding their capabilities, limitations, safety, and robustness. The internship will focus on developing humanoid systems for material handling tasks such as box/cart transportation, kitting, tool manipulation, failure recovery, and other mobile manipulation tasks in warehouse and manufacturing environments. The intern will collaborate closely with MERL researchers to develop novel algorithms that integrate Vision-Language-Action (VLA) models, world models, and robot control for robust long-horizon manipulation on real humanoid platforms. Successful candidates should have demonstrated research experience in embodied AI, robot learning, or computer vision, with the goal of publishing at leading robotics and AI conferences. Start date and duration is flexible.

Required Specific Experience

• Current enrollment in a Ph.D. program in Mechanical Engineering, Computer Science, Electrical Engineering, or a related field.
• Hands-on research experience with Vision-Language-Action (VLA) models and/or multimodal foundation models for robotics.
• Research experience with world models for robotic planning, prediction, or decision making.
• Sim-to-real transfer and robotic simulation (Isaac Lab, MuJoCo.).
• Strong publication record or demonstrated research potential in robotics, embodied AI, computer vision, or machine learning. Must have some publications in top robotic or AI conferences / journals (e.g., RSS, ICRA, T-RO, RA-L, NeurIPS, ICML etc).

Desired Experience

• Contact-rich manipulation and force-controlled manipulation.
• Whole-body control for humanoid robots.
• Dexterous manipulation or bimanual manipulation.
• Reinforcement learning or imitation learning for robotics.
• Robot perception, 3D vision, or state estimation.
• Sim-to-real transfer and robotic simulation (Isaac Lab, MuJoCo).
• Experience deploying algorithms on humanoid platforms.

The pay range for this internship position will be 6-8K per month.