Drone Autonomy.pdf
Summarized by GPT 6.1
Drone Autonomy — Abridged Meeting Notes
October 13, 2026 · Aaron Wu, Shrey Patel, Jorge Ortiz Solano, Soham Goel, Arthur Armaing
Key agreements
- Arthur owns Task 1; Soham owns Task 4. Each owns both the task’s behavior tree and its underlying actions.
- Reuse the boat’s autonomy framework and task templates, adapting topics and behaviors for the drone.
- Task 1 should combine drone surveying with boat guidance: the drone identifies promising search areas, populates the shared map, and responds to boat requests.
- Task 4 must interrupt other tasks and allow the interrupted task to resume afterward. Templates support this pattern; higher-level switching logic still needs integration.
- Keep behavior trees in XML with C++ integration. Complex actions can run as Python ROS 2 action servers.
Near-term priorities
- Target a basic running drone behavior tree by the end of the week.
- For Saturday’s lake test, prioritize reliable building blocks: following the boat, moving to a commanded position or distance, and triggering the payload servo.
- Precision landing and approaching floating pads were additional goals, subject to feasibility.
- Validate actions in simulation before lake testing. Test tree logic separately with generated scenarios.
Action items
| Owner |
Next step |
| Arthur |
Review the boat framework and competition requirements; design and implement Task 1 actions and control flow. |
| Soham |
Review the boat code and begin Task 4 implementation. |
| Arthur + Soham |
Adapt the framework, prepare core actions, and validate them for lake testing. |
| Shrey |
Share visualization setup instructions and the custom-node XML needed to open the trees. |
Open issues
- Confirm the simulation worlds have been pushed and are accessible.
- Convert the existing boat-following implementation into a controllable action server with cancellation support.