All roadmaps
Career pathAdvanced

Spatial AI Engineer

Your destination: Spatial AI Engineer

Teach robots to see, track, and map the world in 3D.

7 courses~90h12 weeks227 lessons
Start this track

About this path

Teach a robot to perceive and reconstruct the world in 3D. Starting from sensor fusion and distributed data plumbing, you'll build up through stereo 3D vision and indoor positioning, visual-inertial odometry, and finally SLAM — from cloning classical papers to neural mapping with NeRF and 3D Gaussian splatting. The path that turns cameras and IMUs into a live map.

Skills you'll gain
Computer VisionSLAMVIOSensor FusionNeRFPythonC++

What you'll be able to do

  • Fuse IMU data into a stable, drift-resistant attitude and motion estimate
  • Recover 3D structure from stereo cameras and build an indoor positioning system
  • Implement visual-inertial odometry, both classical and deep-learning based
  • Build a SLAM system from classical papers up to neural mapping (NeRF, 3D Gaussian)
Before you start

Solid Python and linear algebra; comfortable with C++ basics. IMU Sensor Fusion is a recommended warm-up.

The path7 courses, in order
  1. 1
    Robotics 101 For ProfessionalsBootstrap robotics from first principles — control theory, IMU sensor fusion, and Kalman filtering, all derived and coded in simulation.
    AvailableBeginner · ~28h · 69 lessons
  2. IMU Sensor Fusion Basics for Autonomy EngineersTurn noisy accelerometer and gyro data into a stable attitude estimate.
    PlannedIntermediate · ~10h · 14 lessons
    Planned
  3. Claude Code and Distributed Robotics for EngineersWire up distributed robot nodes with ROS 2, Zenoh, and iceoryx2 — orchestrated with Claude Code.
    PlannedIntermediate · ~14h · 16 lessons
    Planned
  4. 3D Visions for Spatial AI / Robot Perception EngineersRecover 3D structure from stereo cameras and build an indoor positioning system.
    PlannedIntermediate · ~22h · 22 lessons
    Planned
  5. 5
    Visual OdometryEstimate motion without GPS — build feature-based and keyframe visual odometry, then optimize the trajectory graph with g2o on KITTI and EuRoC.
    AvailableIntermediate · ~16h · 62 lessons
  6. Visual-Inertial Odometry for Autonomy EngineerFuse camera and IMU into drift-resistant visual-inertial odometry — classical and deep.
    PlannedAdvanced · ~20h · 20 lessons
    Planned
  7. Classical to Neural SLAM for Advanced Autonomy EngineerGo from classical SLAM papers to neural mapping — NeRF, 3D Gaussian, learned loop closure.
    PlannedAdvanced · ~24h · 24 lessons
    Planned
  8. Spatial AI Engineer · certificateCertified Spatial AI Engineer — built the perception stack from cameras to neural SLAM.