AI is unlocking new opportunities for robotics to handle unstructured tasks -- like complex, high-mix picking in manufacturing facilities.

In this video, we have a conversation with Shawn Patel, Co-Founder & CTO of Almond, about the practical applications of computer vision in robotics and his experience developing custom vision pipelines.

We explore several topics from collecting training data to combining multiple vision models -- including RF-DETR, Grounding DINO, SAM 2, and ViT -- to create highly accurate solutions for complex tasks.

What you'll see in the video:
00:00 Introduction - Vision AI for robotics in manufacturing
01:24 Getting to know Almond - AI robots for manufacturers
02:37 Why vision AI for robotic arms? Adapting to unstructured tasks
05:44 Demo: High mix picking with vision AI
07:30 Behind the scenes: Getting started with vision models
10:40 Combining multiple AI models - Grounding DINO, RF-DETR, SAM 2, ViT
12:49 Collecting training data and training custom models
17:44 Model accuracy for robotic arms
24:04 Conclusion

Join a future live session:
https://roboflow.com/webinar

Learn more about Roboflow
https://roboflow.com/

Get to know Almond:
https://almondbot.com/