FusionCore: ROS 2 sensor fusion that outperforms robot_localization
Details
- External ID
- 47934544
- Source
- HN
- Company
- —
- Product
- FusionCore: ROS 2 sensor fusion that outperforms robot_localization
- Website domain
- github.com
- Launched
- April 28, 2026
- Cohort
- —
- Upvotes
- 11
- Upvotes percentile
- 0.62146529562982
- Tags
- —
- Fetched at
- Sept. 7, 2026, 9:26 p.m.
- Updated at
- Sept. 7, 2026, 9:26 p.m.
Description
I built sensor fusion for a mobile robot and reached for robot_localization like everyone does. After spending too long fighting navsat_transform, UTM zone boundaries, and YAML covariance tuning, I wrote my own.FusionCore is a 22 state UKF that fuses IMU, wheel encoders, and GPS in ECEF directly (no coordinate projection, no extra node). It estimates IMU bias, adapts its noise covariance automatically from the innovation sequence, and gates outliers with a chi squared test on every sensor.I benchmarked it against robot_localization EKF on 6 sequences from the NCLT public dataset (University of Michigan, real robot, real GPS, RTK ground truth). It wins 5 of 6. On the 6th sequence (fall, degraded GPS over a long period) it loses badly. RL UKF diverged to NaN on all six.Configs, methodology, and full reproduce instructions are in the benchmarks/ folder.
Enrichment
- Theme
- autonomous robotics, drones, and sensor hardware
- Vertical
- Horizontal
- Function
- Hardware & robotics
- Audience
- Developer
- AI stance
- Not AI
- Project type
- Commercial product
- Normalized one-liner
- ros 2 sensor fusion for robots
- Manually corrected
- False
Could you build this?
No Implementing a high-performance 22-state Unscented Kalman Filter in C++ for ROS 2 fusing IMU, wheel odometry, and raw GPS in ECEF coordinates requires advanced robotics, state estimation mathematics, and geodesy expertise.
What it would actually take: Building this requires extensive mathematical derivation of non-linear state propagation matrices, quaternion kinematics for 3D orientation, Earth-Centered Earth-Fixed (ECEF) geodesy conversions, and numerical stability algorithms for matrix covariance cholesky decompositions. The stack consists of C++20, Eigen3, and ROS 2 middleware with strict real-time deterministic execution guarantees.
Discussion
2 comments analyzed.
Competitors mentioned: RL (Reinforcement Learning-based system)
Concerns raised: Poor performance on sequences with extended degraded GPS (3x worse on one test case), Adaptive noise estimator overcorrects in low-GPS conditions, Visible trajectory discontinuities from GPS outlier handling differences
Competitors
Other products that read as similar to this one — 27 launches clear the similarity bar, closest 8 shown.
Attention rank: #11 of 28 (itself plus its competitors, highest first — normalized so YC and Product Hunt are compared fairly).
Launched 174 days after the earliest competitor.
- genpark-extended-kalman-filter-robot-localization-skill · github · 2026-09-28 · 7 upvotes · similarity 0.49
- rko_slam · github · 2026-09-11 · 6 upvotes · similarity 0.42
- LiDAR-APR-SCR-Results · github · 2026-09-23 · 16 upvotes · similarity 0.37
- Aerial-autonomy-stack · hn · 2026-06-27 · 5 upvotes · similarity 0.36
- Sowbot · hn · 2026-02-23 · 181 upvotes · similarity 0.35
- genpark-utm-mgrs-coordinate-projection-skill · github · 2026-09-28 · 7 upvotes · similarity 0.35
- nadir-core · github · 2026-09-10 · 47 upvotes · similarity 0.35
- TIO-Former · github · 2026-09-14 · 16 upvotes · similarity 0.35
Other launches for this product
- No other launches for this product.
Same idea, different domain
Nobody's really built a hardware & robotics tool for Fintech yet.