Establish Differential Testing Against an Autoware Reference

未关闭
#752 0 条评论 0 个 reaction 已指派 0 人 在 GitHub 查看

维护者通常 2 天内回复

还没有人认领这个 Issue。

评估

难度
5/5
预计耗时
一周以上
新手友好度
18/100
Issue 类型
功能
描述清晰度
基本清楚
活跃度
活跃
技术栈
cpp, rust

调研方向

The issue names no existing files, tests or entry points, so locate the Awkernel preprocessing (crop box, voxel grid) and NDT code first, and read the pinned Autoware reference's equivalent filters for parameter meaning. This is design-heavy: agree on the tolerance and comparison rules with maintainers before coding. Done means both pipelines run on shared inputs and a deliberately injected mismatch is reported in the machine-readable summary.

由索引模型根据 Issue 内容生成。

描述

Build a reproducible comparison workflow between the Awkernel preprocessing-to-NDT pipeline and a fixed Autoware reference implementation using identical inputs and equivalent algorithm parameters.

Scope:

  • Pin the Autoware commit, relevant dependencies, build configuration, and parameter files
  • Prepare shared test datasets, point cloud maps, coordinate transforms, timestamps, and per-scan initial pose estimates
  • Compare intermediate outputs after crop box filtering and voxel grid downsampling
  • Compare final NDT poses, convergence outcomes, and iteration counts
  • Define numerical tolerances and comparison rules for point ordering, point fields, and floating-point differences
  • Generate machine-readable results and a summary of mismatches
  • Implement Point Cloud Replay and Validate the Preprocessing-to-NDT Pipeline

Acceptance criteria:

  • Both implementations process the same scans, maps, and initial pose estimates without independently evolving their inputs.
  • Parameter mappings and any unavoidable implementation differences are documented.
  • Intermediate point clouds are compared using an order-independent method where ordering is not part of the interface contract.
  • Position and orientation differences are evaluated against predefined tolerances.
  • Convergence-status differences and iteration-count differences are reported; exact iteration-count equality is required only where explicitly justified.
  • The comparison detects deliberately introduced output mismatches.
  • Test data identifiers, commands, configurations, and results are sufficient to reproduce each comparison.

Out of scope:

Performance benchmarking, scheduler evaluation, EKF integration, and proof of correctness against ground truth.

主要语言
Rust
星标
282
派生
12
平均合并
2 天 9 小时
30 天内合并 PR
28

环境准备

这个项目没有提供开发容器、Dockerfile 或贡献指南,环境需要你自己搭建:先看它的 README,通用步骤见我们的新手贡献指南。

从这里开始

  1. 先读完整个 Issue,再读项目的贡献指南。
  2. 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
  3. Fork 仓库,在一个分支上完成修改。
  4. 提交 Pull Request,并在描述里引用这个 Issue 编号。

tier4/awkernel 的其他 Issue

查看 tier4/awkernel 的全部 Issue

相似的 Issue

更多 Rust Issue

把新 issue 发到你的邮箱

精选适合新手参与的 GitHub issue 摘要。