What are the responsibilities and job description for the Deep Learning Based Dexterous Manipulation position at Honda Research Institute USA?
Job Number: P23INT-15
The main objective of this research is to advance machine learning methods, specifically behavior cloning and reinforcement learning, to enable multi-fingered robot hands to dexterously manipulate objects in unstructured human environments utilizing practical sensing (visual and tactile) and actuation components. The candidate is expected to leverage both simulation and real hardware to implement, evaluate, and refine the learned policy.
San Jose, CA
Duration
3 Months
Position Introduction
The main objective of this research is to advance machine learning methods, specifically behavior cloning and reinforcement learning, to enable multi-fingered robot hands to dexterously manipulate objects in unstructured human environments utilizing practical sensing (visual and tactile) and actuation components. The candidate is expected to leverage both simulation and real hardware to implement, evaluate, and refine the learned policy.
Key Responsibilities
During the time of the internship, you are expected to:
- Develop methods to learn sensorimotor policies for dexterous object manipulation using multi-fingered robot hands.
- Implement and validate the developed methods in simulation and on hardware.
- Implement baselines and perform benchmarking to evaluate the learned policy.
- Contribute to the creation and evaluation of various related technologies.
Minimum Qualifications
- Ph.D. or highly qualified M.S. candidate in robotics, computer science, mechanical engineering, or a related field.
- Experience in robot manipulation, control, and planning.
- Experience with building simulated environments (e.g., in Isaac Gym or MuJoCo) and training manipulation policies in simulation.
- Proficient with Python and C .
- Proficient with PyTorch or TensorFlow.
- Experience with Robot Operating System (ROS).
Bonus Qualifications
- Experience with policy deployment on real world robots.
- Experience with representation learning using RGB, RGB-D, and tactile data as inputs.
- Experience with robotic in-hand dexterous manipulation.
- Knowledge in contact dynamics and contact mode switching.
- Experience in online reinforcement learning with hardware robots.
- Experience with Sim2Real approaches.
Years of Work Experience Required:
0
Position Keywords
dexterous manipulation, robotics, deep learning
Warning:You do not have the permission to access the upload fields on this form. Contact the form owner or portal administrator to request the access.
Staff Applied Scientist (Machine Learning)
Deep Abacus -
Santa Clara, CA
Deep Learning Library GPU Software Development Engineer
The Learning Experience #393 -
Santa Clara, CA
Deep Learning Compiler Engineer
VirtualVocations -
Sunnyvale, CA