Advancing the Design of Visual Debugging Tools for Roboticists
summary
Compares a custom AR debugging tool to RViz and a 2D GUI with 24 roboticists (8 per condition) on a 30‑minute detection task and a 1‑hour navigation task, and finds AR can show immersive 3D data but still suffers real‑world issues: occluded overlays hide predicted paths, users still switch to the terminal for errors, and replay of past/future states is rarely used despite demand. Recommends grouping views by task, surfacing frequent streams, enabling dynamic saliency (eg make the robot translucent), and reusing standard color encodings.
AI-written summary · about
chapters (9)
- 0:00Introduction to visual debugging
- 1:13Designing debugging tools for roboticists
- 2:27Categorizing robot data types
- 3:59Visual encoding for robot data
- 4:45Debugging data tasks and AR tool
- 5:54User study on debugging tools
- 6:58Key findings from user feedback
- 8:11Guidelines for future tools
- 9:45Conclusion and future work
related talks
37mThe Bright Future of Debuggers: Challenges and Opportunities
44mShifting gears: How to win with AR
08mAdvancing Mixed Reality Game Development: An Evaluation of a Visual Game Analytics Tool in Action...
08mRapido: Prototyping Interactive AR Experiences through Programming by Demonstration
06m“RustRover in 2025: Code • Debug • AI”
26mTime-awareness in Object-exploration Tools: Toward In Situ Omniscient Debugging
35mVisual Reasoning and Mental Maps in Elixir
16mImproving Communication Between Pair Programmers Using Shared Gaze Awareness