This is a great video. It and your video about Py Trees gave me a much better understanding of both the theory and practice of behavior trees. That said, after doing some more research, I think this video needs a caveat that behavior trees aren't strictly better than FSMs. The speaker at 19:26 of "AI Arborist: Proper Cultivation and Care for Your Behavior Trees" makes a good case that, broadly speaking, AIs should use FSMs to decide _what_ to do, and then use BTs to decide _how_ to do it.
I was using "smach" library in ROS1 and trying to find a better or equivalent in ROS2. However, I couldn't find any quick verbal explanation about alternatives. Thanks a lot, it was very helpful.
Nice explanation! I would like to see more videos on how you decide which behaviors should be clubbed together vs modularized (design decision). Also, Is there a third architecture that we can evaluate (which is neither FSM or behavior tree) to implement the same with similar or even better reactivity?
Thanks Naman! I always feel that design decisions are super subjective and should follow the ideas of “separation of concerns” (component based software engineering). But there is no generic solution (imho) Frankly, I haven’t worked with other task switching mechanisms yet. I did look it up a little after reading this, but I don’t have the working knowledge of other ones
Bro, you're great! Thanks for all the ROS videos. Much appreciated!
Very nice explanation. I was trying to find an overview of behavior trees for robotics and this is by far the best I’ve found.
great job on the videos that actually matter for roboticists. showing a use case and explaining verbally is great.
This is a great video. It and your video about Py Trees gave me a much better understanding of both the theory and practice of behavior trees.
That said, after doing some more research, I think this video needs a caveat that behavior trees aren't strictly better than FSMs. The speaker at 19:26 of "AI Arborist: Proper Cultivation and Care for Your Behavior Trees" makes a good case that, broadly speaking, AIs should use FSMs to decide _what_ to do, and then use BTs to decide _how_ to do it.
I was looking for something like this, mate. Thank you for sharing, and I'm looking forward to the next videos.
Thanks for your constant support Mathias!
Next one coming out this Sunday :)
Great Explanation!!! and thank you for making this video ❤️
I’m glad. Thank you
I was using "smach" library in ROS1 and trying to find a better or equivalent in ROS2. However, I couldn't find any quick verbal explanation about alternatives. Thanks a lot, it was very helpful.
This video really gives me to push in to robotics world ❤❤.
Nice explanation! I would like to see more videos on how you decide which behaviors should be clubbed together vs modularized (design decision). Also, Is there a third architecture that we can evaluate (which is neither FSM or behavior tree) to implement the same with similar or even better reactivity?
Thanks Naman!
I always feel that design decisions are super subjective and should follow the ideas of “separation of concerns” (component based software engineering). But there is no generic solution (imho)
Frankly, I haven’t worked with other task switching mechanisms yet. I did look it up a little after reading this, but I don’t have the working knowledge of other ones
Nice video.
Bro, you're great! Thanks for all the ROS videos. Much appreciated!