trajectory-evaluation

Python package for the evaluation of odometry and SLAM

CI quality gates for SLAM, point clouds, perception, and 3D reconstruction.

Resources for SiTAR, a situated trajectory analysis system for AR which provides in-the-wild pose error estimates

Timestamp Align between the trajs. from VICON & Estimated from Algo.

Cline is an AI coding assistant that keeps a log of what it did. Clinescope reads that log and tells you: whether it used the tools you expected, whether its edit instructions were in the form Cline needs, whether it retyped whole blocks instead of making small changes, and whether it retried the edits that failed.

A world model as a judge: early, auditable verdicts on partial robot-manipulation rollouts

Sanity checks for using small VLM rerankers (Qwen3‑VL 2B) to score candidate driving trajectories from natural‑language instructions.

End-to-end agent evaluation — trajectory eval, tool-use correctness, cost-per-task, latency budgets, regression suites with golden trajectories, LLM-as-judge with calibration. For full agent runs, not just classifiers.

6-DoF ego-motion estimation in CARLA using RGB-D visual odometry — ORB features, PnP-RANSAC, and ATE evaluation

Clean public mirror for a biology tool-use trajectory benchmark with validated Stage A post-training artifacts

Evaluate SLAM trajectory accuracy against ground truth — APE, RPE, Umeyama alignment, and publication-ready plots. No ROS required.

Camera, LiDAR and IMU fault-injection benchmark with timing, synchronization and trajectory diagnostics.

Agentic AI for pharmaceutical supply operations, and the trajectory eval harness that proves it works. Measures the path not just the answer: tool precision, step efficiency, hallucinated-argument rate, unauthorized-write detection, loop/budget controls, and a CI release gate. Zero runtime deps, runs offline.

Build OR optimization models with Openpangu-7B, supporting fine-tuning, automated modeling, constraint generation, and variable pruning