Skip to content
Home / Catalog / Robust Pose-Graph Optimization with False Loop Closures

Robust Pose-Graph Optimization with False Loop Closures

  • Task ID: robotics.robust_pose_graph_false_loops
  • Domain: robotics
  • Subdomain: slam_pose_graph_optimization
  • Status: final
  • Benchmark set: seed42 (60 tasks)
  • Tags: slam, pose_graph, robust_optimization, false_loop_closure, perceptual_aliasing, angle_wrapping

Runtime and requirements

  • Estimated time: 5-25 minutes
  • Python: >=3.10
  • Packages: numpy>=1.24.0, scipy>=1.10.0, matplotlib>=3.7.0
  • GPU required: no
  • Network required: no

Public input and output contract

Inputs

  • poses_initial.npy (data): Odometry-integrated initial trajectory, shape [N, 3] float64 as x_m, y_m, theta_rad.
  • odometry_edges.csv (data): Sequential relative-pose constraints: edge_id, from, to, dx, dy, dtheta, sigma_trans, sigma_rot.
  • loop_candidates.csv (data): Additional non-sequential relative-pose constraints with the same columns. Some candidates are false positives.
  • task_info.json (data): Pose/edge conventions, number of poses and loop candidates, and required output contract.

Outputs

  • analysis.py (code): Agent implementation.
  • poses.npy (data): Optimized 2-D poses [x_m, y_m, theta_rad].
  • loop_closure_scores.csv (data): One row per loop candidate with columns loop_id,false_probability.
  • metrics.json (data): Agent-computed diagnostic metrics and settings.
  • map_diagnostics.png (figure): Diagnostic plot of initial/optimized trajectory and accepted/rejected loop closures.

Public repository files

View this task in ASI-Bench

Formal benchmark task directories in the public repository are metadata-only. Versioned prompts and inputs are distributed through the pinned benchmark dataset.

Provenance

This task is included in the current seed42 benchmark set at dataset revision f11a199f71fb4b854ac43a1bf548d5df519141a9.

This page is generated only from files tracked in the public ASI-Bench repository at commit c9c1ba960854. Submission bundles, run logs, private scoring configuration, and private reference answers are not read by this page generator.