PhD Internship: Hybrid Learning for Humanoid Manipulation

Mitsubishi Electric Research Laboratories, Inc.; Cambridge (MA); On-site; USD 67,000 - 89,000; Part time

Job summary

Mitsubishi Electric Research Laboratories (MERL) invites a Ph.D. student to join a project on hybrid learning and control for reliable humanoid manipulation in factory automation.

You will explore Vision-Language-Action models and World-Action Models, integrating them with motion planning, feedback control, and perception on physical robots. You will implement and evaluate algorithms on real platforms for tasks such as container stacking, kitting, tool manipulation, and assembly, with

Qualifications

• Current enrollment in a Ph.D. program in robotics, CS, ME, EE, or a related field.
• Hands-on experience adapting or evaluating VLA models on physical robots.
• Strong foundations in robot manipulation, motion planning, and feedback control.

Responsibilities

• Develop and evaluate novel algorithms on physical robotic platforms for industrial tasks.
• Combine Vision-Language-Action models with World-Action Models and classical motion planning, control, and perception.
• Compare approaches via task success, recovery performance, execution time, and constraint violations.

Skills

Robot manipulation

Motion planning

Feedback control

Education

PhD student in robotics / CS / ME / EE

Tools

Python

ROS2

PyTorch

Isaac Sim

MuJoCo

Job description

Mitsubishi Electric Research Laboratories (MERL) invites a Ph.D. student to join a project on hybrid learning and control for reliable humanoid manipulation in factory automation.

You will explore Vision-Language-Action models and World-Action Models, integrating them with motion planning, feedback control, and perception on physical robots. You will implement and evaluate algorithms on real platforms for tasks such as container stacking, kitting, tool manipulation, and assembly, with