Robotics Intern - Perception and Navigation

About the Role: You will develop perception and navigation capabilities for a humanoid robot with several onboard cameras: real-time localization and mapping, semantic understanding of industrial equipment, and autonomous navigation in GPS-denied facilities such as refineries, offshore platforms, and processing plants. Simulation will support development and large-scale evaluation before physical robot trials. Build on our SLAM and navigation stack through one focused project. Mapping and navigation work may include persistent maps, relocalization, exploration, planning around measured locomotion limits, and recovery from blocked routes or lost localization. Industrial perception work may include equipment detection, instrument readings with uncertainty, automated annotation, 3D map and operator overlays, and accuracy and latency tests across varied conditions. Evaluate in simulation and validate on robot data. This role offers ownership of a substantial engineering project, with direct technical guidance from the founding team. We value initiative, organization, and persistence. We are flexible about the approach, provided that the solution works reliably on the robot. What You Will Bring

Education: Graduate studies in Robotics, Computer Vision, Machine Learning, or a related engineering field. Simulation experience: Practical experience configuring robotic simulations and using them to develop or evaluate algorithms. Robotics foundations: Knowledge of at least one of multi-view geometry and SLAM, image segmentation and detection, or motion planning and decision-making. Programming: Proficiency in Python or C++, with experience developing software on Linux. Engineering approach: An organized approach to experiments, debugging, and documentation. Ownership: Motivation to learn unfamiliar methods and carry a substantial project through implementation, integration, and evaluation. Useful Additional ExperienceROS 2, Nav2, OpenCV, or PyTorch. Visual or visual-inertial SLAM systems such as ORB-SLAM, VINS-Mono, or Kimera. Camera calibration, multi-camera extrinsics, sensor fusion, or point cloud processing. OCR, analog gauge reading, or 3D semantic mapping. Deployment with real sensor data on physical robots.

Compensation & Benefits: Hourly-based compensation