Research Engineer, Robot Learning

Halodi Robotics; San Carlos (CA); On-site; USD 120,000 - 180,000; Full time

Job summary

Halodi Robotics in California seeks a robotics ML engineer to develop sim2real reinforcement learning (RL) training pipelines for humanoid robots used in in-home settings. You will train and test robot policies via RL, aiming to optimize training pipelines for performance and speed, and model physical systems to close the gap between simulation and hardware.

Responsibilities include collecting and retargeting human motion data for RL pipelines, and supporting testing and debugging on both

Qualifications

• Master’s degree or foreign degree equivalent in Robotics, AI/Machine Learning, Mechanical or Electrical Engineering, Computer Science, or related field
• One year of experience in Robotics & Machine Learning occupation or related in the job offered

Responsibilities

• Develop sim2real RL training pipelines for humanoid robots.
• Train and test robot policies via reinforcement learning.
• Optimize training pipelines for performance and speed.
• Model physical systems including sensors and contacts to reduce sim-to-real gap.
• Collect, clean and retarget human motion data for RL training pipelines.
• Support test and debugging on simulation and hardware.

Skills

Reinforcement learning

Python

C++

Rigid body dynamics

Sim2real

Education

Master's degree

Tools

ROS

Gazebo

Job description

Job Description: Develop sim2real reinforcement learning (RL) training pipelines for humanoid robots. Train and test robot policies via reinforcement learning for in-home humanoid robots. Optimize training pipelines for improved performance and training speed. Modeling of physical systems (including robot dynamics, sensor modeling, contact parameter tuning) to close the gap between simulation and physical systems. Collecting, cleaning and retargeting human motion data for use in RL training pipelines. Supporting the test and debug of RL policies in simulation and on physical hardware.
Requirements: Master’s degree or foreign degree equivalent in Robotics, AI/Machine Learning, Mechanical or Electrical Engineering, Computer Science, or related field and one (1) year of experience in a Robotics & Machine Learning occupation or related in the job offered.
Skills: Experience and/or education must include:

1. Rigid body dynamics and/or dynamical systems for training, deploying and testing reinforcement learned policies for humanoid robots.

2. Implement production-grade Python code for reinforcement learning (RL) and other machine learning training pipelines.

3. Reinforcement learned algorithms, techniques, and sim2real approaches as they pertain to robotics to perform various human-like capabilities, including whole-body loco-manipulation.

4. Implement and debug real-time software in C++ deployed on robots to satisfy various real-time constraints.

5. Model physical systems, including multi-body hybrid dynamic systems with contact, in a simulated environment to identify gaps between simulation and hardware and the ability to understand how to accurately model the hardware in simulation.

6. Root-cause analysis as it pertains to robotic systems and the ability to quickly troubleshoot issues.

7. Forward and inverse kinematics as they pertain to humanoid system training using human motion data.