Robotics Reinforcement Learning Engineer Reinforcement Learning Sim2Real

AsiaCruit, Inc.; New Delhi; Vor Ort; INR 1.800.000 - 2.800.000; Vollzeit

Zusammenfassung

AsiaCruit, Inc. in New Delhi seeks a Robotics Reinforcement Learning Engineer to develop locomotion policies for autonomous quadruped robots using RL and physics-based simulation.

You will train policies to maintain balance, handle disturbances, and adapt across terrains, then validate them on real hardware. You will work with NVIDIA Isaac Lab/Isaac Gym, RaiSim, or MuJoCo, implementing PPO in Python/PyTorch, and deploying models with C++ and ONNX LibTorch.

Qualifikationen

• Bachelors or Masters degree in robotics or related field.
• 3–5 years of RL and autonomous robotics experience.
• Strong Python and PyTorch experience.
• Experience with PPO or similar RL algorithms.
• Experience with robotics simulators (NVIDIA Isaac Lab, Isaac Gym, RaiSim, MuJoCo).
• Understanding of legged robot dynamics and real‑time control.
• Experience deploying models to hardware or edge platforms.

Aufgaben

• Develop RL-based locomotion policies for quadruped robots.
• Train policies across randomized terrains and disturbances.
• Implement domain randomization for robust sim-to-real transfer.
• Improve sim-to-real transfer and policy stability on hardware.
• Deploy policies using C++, ONNX/LibTorch, and related frameworks.
• Collaborate with robotics controls, simulation, and embedded teams.
• Evaluate performance in both simulation and real hardware.

Kenntnisse

Python programming

PyTorch

Reinforcement learning

Problem solving

Ausbildung

Bachelors or Masters in Robotics / CS / EE / ME

Tools

NVIDIA Isaac Lab

Isaac Gym

RaiSim

MuJoCo

C++

ONNX Runtime

LibTorch

Jobbeschreibung

Job Description

About Asiacruit

At Asiacruit, we connect top talent with forward-thinking organizations across industries. Our mission is to help businesses grow through smart, strategic, and people-focused solutions. We support companies by providing high-quality talent for local and global markets.

Position Overview

We are looking for a Robotics Reinforcement Learning Engineer to develop and deploy reinforcement learning-based locomotion policies for autonomous quadruped robots.

In this role, you will train robotic systems to maintain dynamic balance, respond to physical disturbances, and navigate different terrain conditions. You will use physics-based simulation and reinforcement learning to develop robust control policies, then transfer and validate those policies on physical robotic hardware.

The role combines reinforcement learning, legged robot locomotion, simulation engineering, and real-time deployment, with a strong emphasis on improving reliability when moving from simulation to real-world operation.

Key Responsibilities

• Develop reinforcement learning-based locomotion policies using NVIDIA Isaac Lab, Isaac Gym, RaiSim, or comparable simulation platforms.
• Train parallelized locomotion policies across randomized terrain, friction conditions, and dynamic disturbances.
• Develop blind and vision-assisted locomotion approaches for quadruped robots operating in different environments.
• Implement domain randomization across physical parameters, including friction, motor latency, payload variations, and mass distribution.
• Improve Sim-to-Real transfer to ensure learned policies remain stable and effective when deployed on physical robots.
• Implement and optimize reinforcement learning algorithms, including PPO, using Python, PyTorch, and related tools.
• Deploy trained policies using C++, ONNX Runtime, LibTorch, or other suitable inference frameworks.
• Optimize neural policy inference for high-frequency robotic control and low-latency execution.
• Integrate foot-contact estimation, slip detection, and reflex mechanisms to improve stability and reduce the risk of joint damage.
• Test and evaluate locomotion performance in simulation and on physical quadruped platforms.
• Analyze control failures, investigate performance gaps between simulation and hardware, and implement iterative improvements.
• Work closely with robotics controls, simulation, and embedded systems engineers to support integrated robot autonomy.

Qualifications

• Bachelors or Masters degree in Robotics, Computer Science, Electrical Engineering, Mechanical Engineering, Control Systems, or a related field.
• Approximately 3–5 years of relevant experience in reinforcement learning, robot learning, locomotion control, or autonomous robotics.
• Strong Python programming skills and practical experience with PyTorch.
• Experience training reinforcement learning policies, particularly using PPO or similar algorithms.
• Hands‑on experience with NVIDIA Isaac Lab, Isaac Gym, RaiSim, MuJoCo, or other physics-based robotics simulators.
• Understanding of legged robot dynamics, locomotion control, dynamic balance, or contact-rich robotic systems.
• Practical knowledge of domain randomization and Sim-to-Real transfer techniques.
• Experience with C++ and deploying learned models on robotic hardware or edge platforms.
• Familiarity with ONNX Runtime, LibTorch, or comparable inference frameworks is an advantage.
• Knowledge of contact estimation, slip handling, motor dynamics, or real‑time control is a plus.
• Strong problem‑solving skills and the ability to validate reinforcement learning policies through systematic testing and experimentation.

Why Join?

• Work on advanced robotic autonomy and reinforcement learning technologies.
• Develop locomotion policies that enable quadruped robots to handle complex and unpredictable terrain.
• Gain hands‑on exposure to simulation‑to‑hardware deployment and real-world robotics challenges.
• Collaborate with specialists across robot learning, motion control, and embedded systems.
• Contribute to practical R&D projects focused on improving robotic stability, adaptability, and autonomy.

Work Location & Arrangement

On-site at the robotics R&D facility in New Delhi, India