Robotics Perception & State Estimation Engineer - Humanoid (human)

\n\nYour mission & challenges

As a Robotics Perception & State Estimation Engineer, you will develop the core perception capabilities that allow our humanoid robots to understand, localize within, and interact with the physical world.

This role is not focused solely on developing perception algorithms or training machine learning models. We are looking for engineers who can build and own complete perception systems, from sensor integration and calibration through localization, sensor fusion, world modeling, and deployment on real robotic platforms.

You will work at the intersection of perception, state estimation, robotics software infrastructure, and system integration. Your contributions will directly impact navigation, manipulation, safety, and human-robot interaction.

• Own the design, development, and deployment of production-grade perception systems for humanoid robots, transforming raw sensor data into reliable world understanding for navigation, manipulation, and human-robot interaction.

• Architect robust perception software stacks that operate under real-time constraints, from sensor acquisition and calibration to localization, scene understanding, and world modeling.

• Develop and improve state estimation, localization, mapping, SLAM, and sensor fusion systems that enable robots to operate reliably in dynamic real-world environments.

• Integrate, calibrate, and synchronize diverse sensing modalities including cameras, stereo and depth sensors, LiDAR, IMUs, force/torque sensors, and joint encoders into a coherent representation of the robot and its surroundings.

• Design and maintain high-performance ROS2 middleware like and DDS and Zenoh based data pipeline, ensuring reliable communication, timing, synchronization, and coordinate-frame management across the perception stack.

• Build, optimize, and maintain core perception capabilities including:

  • SLAM and localization

  • Sensor fusion and state estimation

  • 3D reconstruction and world modeling

  • Object detection, tracking, and scene understanding

  • Semantic and geometric environment representation

• Evaluate, integrate, and deploy learning-based perception solutions where they provide clear advantages, while balancing robustness, maintainability, and real-time performance.

• Diagnose and resolve complex system-level issues involving calibration, synchronization, localization drift, sensor reliability, communication bottlenecks, and perception performance.

• Develop tools and frameworks for visualization, debugging, validation, and continuous performance monitoring of perception systems.

• Validate and benchmark perception capabilities in simulation and on physical humanoid robots, ensuring seamless transfer from development environments to real-world deployment.

• Collaborate closely with controls, manipulation, planning, and platform teams to define interfaces, requirements, and system architectures that enable reliable robot-wide behavior.

• Drive long-term technical direction for perception architecture, software quality, testing, maintainability, and system scalability.

What we can look forward to

• Bachelor's or Master's degree in Computer Science, Software Engineering, Robotics, or a related technical field.

• 3+ years of professional experience building and deploying complex robotic perception systems, not just individual perception algorithms or ML models.

• Proven ownership of key components of a perception stack, such as sensor integration, calibration, localization, mapping, sensor fusion, state estimation, or real-world deployment.

• Hands-on experience with one or more of the following:

  • Visual, LiDAR, or Visual-Inertial SLAM

  • Localization and mapping

  • Sensor fusion and state estimation

  • Factor graph optimization

  • EKF/UKF-based systems

• Strong understanding of robotics software architectures, including:

  • ROS2 and DDS

  • Distributed robotic systems

  • High-throughput sensor pipelines

  • Synchronization, timing, and coordinate frame management (TF)

• Practical experience integrating and debugging perception sensors, including cameras, LiDARs, IMUs, force/torque sensors, and encoders.

• Strong software engineering skills with modern C++ (C++17/20), multi-threaded systems, performance optimization, and production-quality code development.

Nice to have

• Perception for humanoids, manipulation, or autonomous robots

• GPU acceleration

• Embedded robotics systems

• NVIDIA Jetson platforms

• Deep learning for perception

• PyTorch or TensorFlow

• PCL, OpenCV

• IsaacSim, Mojuko or Gazebo

• MLOps experience