Visual Odometry

AI-powered visual odometry enables robust navigation without reliable GNSS signals

Business impact

• Operational efficiency — Enhances task completion speed by maintaining navigation accuracy despite GNSS loss
• Autonomy level — Enables autonomous operation by providing reliable spatial orientation without external signals
• Reliability — Reduces navigation failures caused by weak or disrupted GNSS signals
• Downtime reduction — Minimizes operational interruptions due to navigation system failures in challenging terrain

Data requirements

• Stereo camera images (Image) — Capture sequential visual data to estimate motion and spatial changes
• IMU sensor data (Numeric) — Provide inertial measurements to complement visual odometry for pose estimation
• GNSS signals (Numeric) — Used when available for precise positioning and as a fallback reference

AI methods and techniques

• Predictive AI — Predicts camera motion and pose changes from sequential image frames for localization
• Symbolic AI — Utilizes geometric constraints and feature matching for pose graph optimization

AI models and model families

OpenCV StereoBM, RTAB-Map, Custom CNNs for feature extraction, ORB feature descriptor

Industries

Real-world evidence

1 documented case study on record.

Companies using this: Roboton.