Skip to content

lidarslam_ros2 v0.11.0

The first published release since v0.10.0 of the supported Humble and Jazzy rosbag2-to-Autoware-compatible map workflow. The v0.10.1 candidate was never tagged; its changes ship in v0.11.0.

v0.11.0 is not v1.0. One of its five blocking accuracy profiles, newer_college_math_hard, was not measured for this release; see Release verification.

Shortest supported paths

These are unchanged from v0.10.0:

  • lidarslam-map doctor, then lidarslam-map demo <new-output-dir> for the fixed public MID-360 demo;
  • lidarslam-map doctor /path/to/rosbag2, then lidarslam-map start /path/to/rosbag2 for your own bag.

The release workflow publishes v0.11.0-humble and v0.11.0-jazzy images built from this tag. Getting Started keeps the audited v0.10.0 images until their publication audit.

Changes since v0.10.0

  • Several runs of one site, one map. graph_slam_multisession_join joins separate runs into one pose graph with no position hint. It refuses ambiguous sets and self-similar places rather than guess. It needs small_gicp. See Multi-session join.
  • Loop closure no longer drops input. The loop search runs on a worker thread, so the backend builds every submap and repeats its output from run to run.
  • The backend places RKO-LIO's scan with its LiDAR-to-base extrinsic. With a non-identity extrinsic (HILTI, for example), maps and trajectories change.
  • Default runs repeat. The offline node waits for graph_based_slam before it publishes.
  • Less memory and a faster backend, same outputs. Stored scans are packed and compressed, large maps are saved without assembling them, and the final map refinement is 2.4 times faster. Outputs are byte-identical.
  • --motion walking for the MID-360 preset, with the frontend and loop settings adopted on humanoid walking data.
  • Large maps are downsampled again. A map whose extent overflowed PCL VoxelGrid's voxel index used to be saved with every point.
  • Official RKO-LIO 0.4.1 and later (ROS apt ships 0.4.2) save a map again.
  • lidarslam-map works on the session directories start creates (edit, view, inspect), survives a slow first start, and checks free space before writing.
  • Optional backend features: SMALL_GICP loop registration, a loop drift gate, fixed loop-edge information, and a causal trajectory output in graph_slam_offline_runner. All are off by default.
  • The first independent first-map validation is accepted (1 of the 3 that v1.0 needs).

With loop closure on, or with a non-identity LiDAR extrinsic, results differ from v0.10.0. The full list is in CHANGELOG.md.

Release verification

All gates ran on the clean carrier commit 520c597287522b41266eb50be374c4b4894b6878 with a Jazzy build of that commit (lidarslam-map --version: lidarslam_ros2 0.11.0). The develop merge 8a5d66f5 has the same tree. The tagged commit differs from it only in this section, the changelog heading, the README link, a recorded exception in RELEASING.md and the release-gate snapshot of Benchmarking.

Gate Command Result
Python product suites bash scripts/run_product_python_tests.sh graph_based_slam 2,279 passed / 58 skipped / 46 failed; lidarslam 1,219 passed / 1 failed. The 46 are the competitive-benchmark contract tests the default CI gate excludes; they fail the same way at v0.10.0 on this host. The 1 needs the ip command, which this host lacks.
Strict documentation python3 -m mkdocs build --strict PASS
First-map verification package python3 scripts/check_first_map_verification_package.py --json READY
v1 readiness contract python3 scripts/check_v1_readiness.py --json valid; NOT_READY 8/10 (distribution; independent adoption 1/3)
Release bundle reproducibility python3 scripts/check_release_bundle_reproducibility.py PASS: two byte-identical builds, 18,453,562 bytes, SHA-256 9a647a220551765fda4ece39b89d183c51d90432bb4ea87efd990e20854c0513
Installation check lidarslam-map doctor PASS
Fixed first-map demo lidarslam-map demo <new-dir> --viewer none PASS: map verifier PASS, first-map receipt PASS (7/7)
Default workflow (build and package tests) GitHub Actions ci on Humble and Jazzy PASS on 8a5d66f5 (run 41405565240)
Blocking release profiles bash scripts/run_release_readiness_checks.sh --skip-default-ci --benchmark-root <root> --fail-on-profiles exit 2: NO_DATA for newer_college_math_hard only (exception below)

Blocking release profiles at the carrier commit, each with clean provenance for that exact commit:

Profile Metric v0.11.0 v0.10.0 Pass Target Status
newer_college_math_hard APE RMSE vs ground truth not measured 0.070 m 0.10 m 0.08 m NO_DATA
ntu_viral_tnp_01 APE RMSE vs ground truth 0.995 m 0.975 m 1.00 m 0.30 m PASS
mid360_gt_rtkslam_construction_seq2 APE RMSE vs checkpoints 0.165 m 0.165 m 0.30 m 0.15 m PASS
mid360_gt_rtkslam_construction_seq1 APE RMSE vs checkpoints 0.280 m 0.280 m 0.55 m 0.30 m TARGET_MET
leo_drive_applanix_velodyne_cross APE RMSE vs Applanix 0.139 m 0.139 m 1.50 m 0.50 m TARGET_MET

Exception: newer_college_math_hard was not measured

Google Drive answered "Quota exceeded" for the 9.5 GB part of the official Maths-Hard recording on every attempt on 2026-11-02 (last check 23:38 UTC). The maintainer chose to release v0.11.0 without this profile, as a one-time exception recorded in RELEASING.md. The profile stays blocking for later releases.

There is no Maths-Hard accuracy number for v0.11.0. The v0.10.0 value does not carry over: this release changes results with loop closure on and with a non-identity LiDAR extrinsic, and the local Maths-Hard settings were rebuilt from the dataset's published calibration for this release.

Notes

  • NTU VIRAL. 0.995 m is within the 1.00 m threshold by 5 mm. The raw odometry matches v0.10.0 (0.985 m against 0.984 m); the graph accepted 11 loop closures against 8. v0.10.0 already varied by about 1 cm between runs.
  • NTU acquisition. The official archive was read as a stream and checked against its published size and MD5; only the three topics the benchmark uses were kept, then the download script's own conversion steps ran. A fresh NTU VIRAL preparation still starts with the write- and network-free bash scripts/download_ntu_viral_tnp01.sh --dry-run.
  • Leo Drive. The Velodyne driver overlay was built at upstream revision 56fc178d, the revision of the July evidence. The current upstream tip 8cdc0f37 gives a two-vertex graph on this bag. scripts/prepare_velodyne_pointcloud_overlay.sh does not pin a revision.