In this role, you will develop high-fidelity simulation platforms essential for Apptronik’s general-purpose robots, with the primary objective of enabling seamless transition of control algorithms and autonomous behaviors from simulation to the real world. You’ll ensure seamless integration of simulation with control systems, aligning data flows between virtual and real hardware for accurate system identification, contact modeling, and reinforcement learning. By bridging the sim-to-real gap—through parameter transfer, realistic contact dynamics, and efficient parallelization for reinforcement learning—you will help enhance robotic performance and adaptability.

Responsibilities:

• Full Ownership of Simulation Infrastructure: Develop and maintain simulation infrastructure integrated with robot control systems, ensuring that data flows accurately reflect both simulated and real-world operations.
• System Identification and Sim-to-Real Transfer: Take responsibility for transferring identified robot and actuator parameters into simulation models, ensuring high-fidelity sim-to-real alignment for various robotic systems.
• Realistic Contact Modeling: Design and implement accurate contact models, enabling simulations that match real-world physics and interaction dynamics.
• Scalable Reinforcement Learning Support: Spearhead the development of a simulation platform for reinforcement learning, enabling robust, parallelized training environments across multiple simulations (e.g., Isaac Gym, MuJoCo MJX).
• Photo-realistic Simulation for Perception: Develop visually accurate simulation environments to support perception-related tasks, including photorealism for improved visual model training.
• Experimental Design and Validation: Design and execute experimental procedures to characterize kinematics, dynamics, and controls, ensuring that system identification matches hardware requirements.
• Deliver reliable software through code reviews, continuous integration, and automated testing
• Represent and lead with Apptronik's core values: CHIPC - Curiosity, Humility, Integrity, Passion, Creativity

 

Skills and Requirements:

• Education: BS, MS, or PhD in relevant fields (Computer Science, Robotics, Electrical Engineering, etc.)
• Experience: Significant experience (8+ years or 5+ with a graduate degree) in developing robotic simulations and conducting system identification, particularly in high-performance, dynamic robots.
• Technical Skills: Strong command of modern C++ and object-oriented programming, including expertise with templates, inheritance, and design patterns for robust software design.
• Thorough knowledge of kinematics, dynamics, controls, estimation, and familiarity with standard equipment (e.g., motion capture, force plates, dynamometers).
• Nice to have: Skilled in ROS, URDF, MuJoCo, Isaac Gym, and Eigen, with experience tuning simulations for accuracy and fidelity.
• Nice to have: Experience with designing and optimizing parallelized simulation environments for reinforcement learning applications.
• Strong experience with Linux application development, software testing (unit & integration), and version control (e.g., Git).
• Proven ability to communicate technical information clearly and work within Agile development processes (sprint planning, design/code reviews).