Skip to content

M1: Point Cloud Preprocessing and NDT Localization #745

Open
0 of 5 issues completed
Open

Description

@ytakano

Port and integrate point cloud preprocessing and NDT scan matching into Awkernel to establish a baseline for real-time performance evaluation.

Scope:

  • Crop box filtering and voxel grid downsampling
  • NDT scan matching with a preloaded point cloud map and supplied initial pose estimates
  • Replay of recorded or synthetic point cloud data
  • Comparison against a fixed Autoware reference implementation using identical inputs and parameters
  • Measurement of localization accuracy, execution time, end-to-end latency, and memory usage

Completion criteria:

  • The preprocessing-to-NDT pipeline runs end to end on Awkernel.
  • Output differences satisfy predefined numerical and localization tolerances.
  • Evaluation datasets, configurations, and reproduction steps are documented.
  • Baseline performance results are recorded for subsequent scheduling and integration experiments.

Out of scope:

Gyro odometry, EKF integration, motion distortion correction, live sensor drivers, and dynamic map loading.

Metadata

Metadata

Labels

No labels
No labels

Type

No type

Projects

No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions