Robot Runtime Observability
How can a robot run retain its state, versions, environment, and task context?
Connect topics, TF, logs, parameters, maps, and task events into run records that can be queried and compared.
Research agenda
A long-term research program for robot runtime, diagnosis, and reliability.
How can robot missions be recorded, failures replayed, diagnoses linked to evidence, and engineering improvements verified?
Failure Replay & Diagnosis
How can fragmented data reconstruct a failure timeline without losing the original evidence?
Study reproducible and explainable incident review through Episodes, Incidents, and Evidence.
Navigation Reliability
How do TF, odometry, localization, planning, and control combine to produce navigation failures?
Build a body of SLAM, Nav2, and sensor-chain failure cases around ROS 2 differential-drive robots.
Research method
Research must return to verifiable engineering results.
Reproducible systems
Begin with simulation and progressively introduce recorded runs.
Complete run evidence
Preserve rosbag2, logs, parameters, maps, and software versions.
Episode reconstruction
Rebuild a mission timeline and its operating context.
Diagnosis and improvement
Trace conclusions to evidence and validate fixes through repeated experiments.
Research object
Study one mobile-robot loop deeply.
The current scope is a ROS 2 differential-drive robot across sensors, TF, odometry, SLAM, localization, planning, and control. A focused boundary enables comparable data and failure cases.
View experiments