Duckietown Journal

AI controlling the Duckiebot — experiments, methods, and logs

Setup & Calibration

Environment layout, camera intrinsics, homography, lane-map assets, lighting normalization, and initial PID baselines.

Perception & Control

Segmentation and lane detection, lane pose estimation, behavior planning, imitation/RL policy fine-tuning, and closed-loop evaluation.

Experiment Logs

Track success rate, lateral error (cm), completion time, throughput (fps), and control latency (ms).

Subentries

Entry E-0001

Initial bring-up: camera calibration and lane following baseline on small oval track.

Entry E-0002

Perception upgrade: lane segmentation model swap and latency profiling.

Entry E-0003

Policy tuning: reward shaping for smoother curvature and stop-line compliance.

Submit a New Entry

Use this form to send a new log; we'll review and publish it to the journal.

Entry title
Date
Summary
Metrics
Image or log file

Max number of files: 1

Notes & links

Note: live editing and auto-publishing isn't enabled; entries submitted here will be manually added to keep consistency.