Scientist: Deep Reinforcement Learning for Dexterous Object Manipulation - Scientist: Deep Reinforcement Learning for Dexterous Object Manipulation - HRI-US
Scientist: Deep Reinforcement Learning for Dexterous Object Manipulation
Job Number: P20T04
Honda Research Institute USA (HRI-US) in San Jose, California, is looking for a postdoctoral scientist to investigate sample-efficient deep reinforcement learning approaches to dexterous manipulation of objects. Research topics include incorporation of human expertise through imitation learning, learning from videos, and primitive skill reuse through hierarchical reinforcement learning.
San Jose, CA
- Development of deep reinforcement learning algorithms for dexterous manipulation.
- Implementing the algorithms in simulation and on multi-fingered robot hand hardware.
- Collaborate with local and international researchers and engineers within the company.
- File patents on developed technologies.
- Publish research results at top-tier conferences and journals in robotics as well as machine learning.
- Ph.D. in computer science, robotics, or a related field.
- Experience with deep learning reinforcement learning.
- Familiarity with deep learning platforms such as TensorFlow and PyTorch.
- Experience with simulation engines such as MuJoCo and PyBullet.
- Familiarity with ROS and developing end-to-end systems.
- Experience with deep reinforcement learning deployed on a robotics platform is a plus.
- Experience with robotic manipulation, control, and dynamics.
- Experience with sim-to-real approaches is a plus.