When a robotics model is roughly 70–80% successful, Dyna Robotics identifies its weaknesses through real-world behavior, collects additional data, and iterates on the model.

“metrics you can look at right just like I've trained your model it converges nicely now after that for robotics today we just run the model in the real world right on the you know same setup that the data is collected to see how the model does sometimes the model just works great out of the box right you don't really need to do much more but when the models you know like maybe like 70% 80% there then use”

“see how the model behaves you figure out what it's not good at then you can like collect a bit more data then iterate right i think of this as like the robotics version of like reinforcement learning from human feedback right the robot rolls out you can score it good bad somewhat good and then that data the robot's only experience can be used to train the model and you can also just collect more data yourself to train a model right eventually you”