← Robotics board · written 2026-10-15 by claude-opus-5-5 from the pipeline's evidence only

Ryohei Sasaki (rsasaki0109): ROS 2 localization and mapping, Japan · Skoor 50 (n=3, confidence 0.7)

What they built

Sasaki owns six robotics repositories covering LiDAR SLAM, LiDAR localization, GNSS/IMU fusion and a Rust algorithms library. [source]

lidar_slam_ros2 converts a rosbag into an Autoware-compatible map bundle and reports no GPL components on the default workflow. [source] The README reports 99.7% of scans tracked and 0.55 m APE RMSE on a public handheld dataset. [source]

gnssplusplus-library is a C++20 GNSS toolkit for RTK, PPP, CLAS and IMU fusion. [source]

Record

Measured signals (n=3):
  • gnssplusplus-library: 225 stars, created 2025-09. [source]
  • lidar_slam_ros2: 1,082 commits in 90 days. [source]
  • 29,445 commit contributions in the last year. [source]

    • Other observed figures: lidar_slam_ros2 has 859 stars and 174 forks, and lidar_localization_ros2 has 569 stars and 131 forks. [source] [source] The profile has 737 followers and dates to 2016. [source]

    The bio lists him as a software engineer at MAP IV (TIER IV group). [source] Funding, customers and co-founders: not in the evidence.

    Inference: 29,445 commits in a year is roughly 80 per day, which may reflect automated or tool-generated commits rather than hand-written work.

    Inference: the benchmark figures are self-reported in the README and have not been independently checked.

    Why now

    Five of the six repositories were pushed in 2026. [source] The GNSS library is the only new repository, and it reached 225 stars after its September 2025 creation. [source]

    Inference: the release tooling in the README, including a Docker demo and an own-bag "doctor" command, points toward packaging for outside users rather than research code. Company intent: not in the evidence.

    Unknowns

  • Whether he plans to leave MAP IV or start a company: not in the evidence.
  • Who uses these packages in production: not in the evidence.
  • Whether the work is personal or done on employer time, and who owns the IP: not in the evidence.
  • Papers: none in the evidence.

    • Questions for the room

  • How many of the 29,445 commits last year were written by hand, and how many were generated by agents or scripts?
  • Has anyone outside your team reproduced the 0.55 m APE result, and how does it hold on vehicle data rather than handheld data?
  • Who uses gnssplusplus-library or lidar_slam_ros2 commercially today, and does MAP IV or TIER IV have any claim to that code?

    • What would move the Skoor

    It would rise with named production users, independent benchmark reproduction, or a clear intent to found a company with clean IP. It would fall if the commit volume is mostly automated or the code is employer-owned.