Sprints
Sprint spotlights for the Vision area, generated every week from the project board and reviewed by the area PM before merging. Newest sprint first.
Status legend:
- 💻 Development
- 🔍 Research
- 📝 Documentation
- 🔧 Bug fixing
Sprint 5 · 2026-11-02 → 2026-11-22
Sprint plan
7 tasks planned for this sprint.
Done:
- Arturo Cabrera 💻 Run zed without router (#1453)
Review:
- Fernando Hernandez, Angela Camila Tite Haro, Marco Galindo 🔍 Semantic MAP (#1268)
In Progress:
- Mario Casas Donjuan 💻 Integrate talk detection into follow face (#1431)
- Marco Galindo 💻 Run embedding gallery in its own detector node (#1449)
- Fernando Hernandez 💻 Research LFM2.5-VL por pose detection (#1472)
Todo:
Sprint 4 · 2026-10-12 → 2026-11-01
Sprint results
7/7 tasks done (100%).
Sprint 4 focused on updating TDP documentation and cleaning up the vision environment. The team successfully delivered the InsightFace face recognition replacement, person tracking updates, YOLO11 pose text, embedding gallery contributions, object detector pipeline updates, Docker-l4t dependency cleanup, and CLIP embeddings for semantic navigation. There are no pending tasks remaining.
Done:
- Arturo Cabrera 💻 TDP: Replace DLib face recognition text with InsightFace + TensorRT (#1367)
- Marco Galindo 💻 TDP: Verify and update the person tracking and Re-ID description (#1368)
- Juan Pablo Gutierrez 💻 TDP: Update pose and gesture text (YOLO11 pose, MediaPipe, Moondream2 over gRPC) (#1369)
- Marco Galindo 💻 TDP: embedding gallery contribution (#1370)
- Fernando Hernandez 💻 TDP: update Object detector pipeline (#1371)
- Arturo Cabrera 🔧 Clean docker-l4t (#1377)
- Juan Pablo Gutierrez 💻 Finals commands node: CLIP embeddings for semantic navigation (#1378)
Sprint 3 · 2026-09-21 → 2026-10-11
Sprint plan
5 tasks planned for this sprint.
In Progress:
- Juan Pablo Gutierrez 💻 Detection of the person speaking (P1, #1267)
Todo:
- Fernando Hernandez, Angela Camila Tite Haro, Marco Alejandro Galindo de la Cruz 🔍 Semantic MAP (#1268)
- Fernando Hernandez, Marco Alejandro Galindo de la Cruz 🔍 Vision: embedding-based object recognition for setup-day onboarding (#1310)
- Arturo Cabrera 🔍 Benchmark newer VLM, and general models options (#1311)
- Luis Alvaro Rosales Salazar 🔍 Garment perception for the Doing Laundry (#1312)
Week 1 · 2026-09-21 → 2026-10-02
Review:
- Fernando Hernandez, Angela Camila Tite Haro, Marco Alejandro Galindo de la Cruz 🔍 Semantic MAP (#1268)
- Arturo Cabrera 🔍 Benchmark newer VLM, and general models options (#1311)
In Progress:
- Juan Pablo Gutierrez 💻 Detection of the person speaking (P1, #1267)
- Fernando Hernandez, Marco Alejandro Galindo de la Cruz 🔍 Vision: embedding-based object recognition for setup-day onboarding (#1310)
Not started: 1 task.
Week 2 · 2026-10-03 → 2026-10-09
Done this week:
- Juan Pablo Gutierrez 💻 Detection of the person speaking (P1, #1267)
- Fernando Hernandez, Marco Alejandro Galindo de la Cruz 🔍 Vision: embedding-based object recognition for setup-day onboarding (#1310)
- Arturo Cabrera 🔍 Benchmark newer VLM, and general models options (#1311)
Review:
- Fernando Hernandez, Angela Camila Tite Haro, Marco Alejandro Galindo de la Cruz 🔍 Semantic MAP (#1268)
In Progress:
- Arturo Cabrera 💻 Add text embedding support to ImageEmbedder (#1348)
Updates:
- Arturo Cabrera 🔍 Benchmark newer VLM, and general models options (#1311): Moondream 3.1 evaluation on Jetson AGX Orin was closed without implementation due to a persistent CUDA kernel error during initialization.
- Arturo Cabrera 💻 Add text embedding support to ImageEmbedder (#1348): ImageEmbedder text embedding implementation was reduced to a minimal commit in PR 1358 with passing tests and benchmark results.
Not started: 1 task.
Sprint 2 · 2026-08-31 → 2026-09-20
Sprint plan
4 tasks planned for this sprint.
Done:
- Juan Pablo Gutierrez 💻 Move /utils folder from vision_general to script (P1, #1241)
- Juan Pablo Gutierrez, Arturo Cabrera 💻 Testing zed2 in Jazzy - ubuntu24 (P0, #1243)
In Progress:
- Marco Alejandro Galindo de la Cruz 💻 Migrate vision from humble to jazzy (P0, #1242)
Todo:
- Luis Alvaro Rosales Salazar 💻 Detection of the person speaking (P1, #1267)
Week 1 · 2026-08-31 → 2026-09-06
Done this week:
- Juan Pablo Gutierrez 💻 Move /utils folder from vision_general to script (P1, #1241)
In Progress:
- Marco Alejandro Galindo de la Cruz 💻 Migrate vision from humble to jazzy (P0, #1242)
Not started: 1 task.
Sprint results
5/6 tasks done (83%) · P0: 2/2.
This sprint focused on migrating the vision system to ROS 2 Jazzy and improving CPU-CUDA images. The team successfully migrated the vision system, tested the zed2 camera on Ubuntu 24, and handled utilities, cpu-cuda images, and the vision/hric issue. However, the detection of the person speaking remains pending.
Done:
- Juan Pablo Gutierrez 💻 Move /utils folder from vision_general to script (P1, #1241)
- Marco Alejandro Galindo de la Cruz 💻 Migrate vision from humble to jazzy (P0, #1242)
- Fernando Hernandez, Marco Alejandro Galindo de la Cruz 💻 Testing zed2 in Jazzy - ubuntu24 (P0, #1243)
- Juan Pablo Gutierrez 💻 Improve and migrate cpu-cuda images (#1301)
- Marco Alejandro Galindo de la Cruz 💻 Vision/hric issue (#1309)
Carried over:
- Juan Pablo Gutierrez 💻 Detection of the person speaking (In Progress, P1, moved to Sprint 3, #1267)
Sprint 1 · 2026-08-10 → 2026-08-30
Sprint results
8/8 tasks done (100%).
Done:
- Fernando 💻 Improve Architecture (P1, #411)
- Fernando 💻 Package re-structure (P1, #428)
- Marco Alejandro Galindo de la Cruz 💻 Clear Moondream package (P1, #1135)
- Juan Pablo Gutierrez 💻 Clear Vision subtask and test_vision_manager.py (P1, #1136)
- Marco Alejandro Galindo de la Cruz 💻 General clean (P1, #1138)
- Fernando 💻 Run vision in cpu-cuda (P1, #1139)
- Fernando 💻 Update vision readme (P1, #1140)
- Juan Pablo Gutierrez 💻 GPSR tasks testing (P1, #1177)