Light-Weight Pointcloud Representation with Sparse Gaussian Process

This paper presents a framework to represent high-fidelity pointcloud sensor observations for efficient communication and storage. The proposed approach exploits Sparse Gaussian Process to encode pointcloud into a compact form. Our approach represents both the free space and the occupied space using only one model (one 2D Sparse Gaussian Process) instead of the existing two-model framework (two 3D Gaussian Mixture Models). We achieve this by proposing a variance-based sampling technique that effectively discriminates between the free and occupied space. The new representation requires less memory footprint and can be transmitted across limited-bandwidth communication channels. The framework is extensively evaluated in simulation and it is also demonstrated using a real mobile robot equipped with a 3D LiDAR. Our method results in a 70~100 times reduction in the communication rate compared to sending the raw pointcloud. We have provided a demonstration video 1 1 Video: https://youtu.be/BQZzXiCFGrM and open-sourced our code 2 2 Code: https://github.com/mahmoud-a-ali/vsgp_pcl.

Authors

• Mahmoud Ali
• Lantao Liu

Keywords

• Solid modeling
• Three-dimensional displays
• Laser radar
• Memory management
• Collaboration
• Communication channels
• Robot sensing systems
• Gaussian Process
• Sparse Gaussian Process
• Free Space
• Gaussian Mixture Model
• Compact Form
• Communication Rate
• Raw Point Cloud
• Occupied Space
• Sensor Observations
• 3D LiDAR
• Root Mean Square Error
• Computational Complexity
• Simulation Experiments
• Mean Function
• Part Of Surface
• Elevation Angle
• Covariance Function
• Hyperparameter Values
• True Posterior
• Hardware Experiments
• Original Point Cloud
• Occupancy Map
• Point Cloud Data
• Memory Reduction
• 3D Point Cloud
• Occupancy Values
• Radius Of Point
• Gaussian Density