- 6
- 33 748
Poyen Hsieh
Приєднався 15 гру 2012
Hi! I'm Poyen. I transitioned from civil engineering to architecture, then AI engineering, and now, I'm a happiness coach. The journey across various fields was challenging but rewarding, showing me the importance of not settling for a conventional path. Deep down, I've always sought meaningful pursuits, realizing that true satisfaction transcends mere success. My epiphany came five years ago during a leadership workshop, where I discovered the inner game beneath our everyday goals-a game that dictates all others. With guidance from mentors and coaches, I embraced a shift from finding happiness in success to Happiness Driven Success. This journey unveiled the possibility of experiencing internal happiness now, without waiting for external achievements. I've learned that life is an inside-out phenomenon, shaped by our minds. My aim is to share these insights in an understandable and accessible way, and facilitate meaningful discussions around that with you! Happy exploring!
AI Helps Designers Design Better Kitchen Faster
Creator's website: www.poyenhsieh.com/
Cornell Master of Architecture Thesis, Dec 2019
This paper, titled "Co-Design with DRL Agent in Kitchen Layout," aims at exploring new possibilities of human-machine co-design by utilizing a deep reinforcement learning framework. As the technology matures and the industry’s demand for design automation rises, how and what to automate becomes the critical first step. The following work describes a step-by-step, conversational collaboration with a virtual assistant, in contrast to the conventional ‘solutions-to-select-from’ type of optimization. The benefit here is three-fold: (1) the flexibility to discover any desired computer-rated sub-optimal solution, (2) the intuitiveness to collaborate like a conversation, (3) the authorship retained for human designers. Architects traditionally take decades to train on both reasoning and sensibility, yet machines can now master the former, which creates the perfect aspect to automate. Starting from the functional goals that are more quantifiable and consensual, the virtual assistant can be trained to become the extension of a designer’s mind. To validate this framework, kitchen layout is chosen as the subject for its smaller scale and rich functional requirements. The proposed method contains four main stages: building an environment, collecting data from a self-play mechanism, training a deep neural network, and building a human-machine co-design interface. The current research establishes an alternative design methodology that is novel in approach, borrowing concepts from reinforcement learning, rule-based system, and generative design for the conversational human-machine co-design workflow.
Cornell Master of Architecture Thesis, Dec 2019
This paper, titled "Co-Design with DRL Agent in Kitchen Layout," aims at exploring new possibilities of human-machine co-design by utilizing a deep reinforcement learning framework. As the technology matures and the industry’s demand for design automation rises, how and what to automate becomes the critical first step. The following work describes a step-by-step, conversational collaboration with a virtual assistant, in contrast to the conventional ‘solutions-to-select-from’ type of optimization. The benefit here is three-fold: (1) the flexibility to discover any desired computer-rated sub-optimal solution, (2) the intuitiveness to collaborate like a conversation, (3) the authorship retained for human designers. Architects traditionally take decades to train on both reasoning and sensibility, yet machines can now master the former, which creates the perfect aspect to automate. Starting from the functional goals that are more quantifiable and consensual, the virtual assistant can be trained to become the extension of a designer’s mind. To validate this framework, kitchen layout is chosen as the subject for its smaller scale and rich functional requirements. The proposed method contains four main stages: building an environment, collecting data from a self-play mechanism, training a deep neural network, and building a human-machine co-design interface. The current research establishes an alternative design methodology that is novel in approach, borrowing concepts from reinforcement learning, rule-based system, and generative design for the conversational human-machine co-design workflow.
Переглядів: 1 889
Відео
Generative Design: Kitchen Layout Automation
Переглядів 21 тис.5 років тому
Computational Design Project at CBT Architects, 2019 summer Creator's Website: www.poyenhsieh.com
Grasshopper vs Dynamo in Kitchen Layout Automation
Переглядів 7 тис.5 років тому
Grasshopper Galapagos vs Dynamo Refinery. Find out who wins in the fair battle of kitchen layout automation with identical python script and customized visualization for each software. The result only represent comparison in efficiency of solvers, Galapagos and Refinery, in this particular project. Creator's Website: www.poyenhsieh.com
On Robotic Weaving
Переглядів 8005 років тому
Matter Design Computation at Cornell University with Jenny Sabin Groupwork: Poyen Hsieh, Hafsa Muhammad, Ellie Krause Creator's Website: www.poyenhsieh.com
Hahaha... I hope I can always remember that. I am NOT my mind's programing. Great video!
To what extent does this analogy stand true for you? If it applies, what do you want to uninstall?? 😁
Yeah! I want to uninstall fear that stop me from trying new things!
That's a good one :D@@lllwertyu
i like how their are many dark ways this can be taken via sugar coding, like "i wanna hit the power off button"💀💀
It can definitely improve. Since is not the program but the compilator, if your compilator sucks you wont get the best of the programs you're trying to install. This new age thinking of, everything has to be good all the time or profits always have to go up is whats really making the journey harder than it should. Feeling sadness, sorrow or anger is ok as long as you are in control of it. Is not about uninstalling programs or just rejecting what you dont like, but having a really good OS that upgrades and stands its ground when things go south. Our bodies are made to grow better through struggle. Thats just how carbon based life has behave for over... 4M years?
Can you recall a specific experience that led you to believe success brings happiness? I’m curious to know! 😉
Yea It's obvious that when I love what I do, I do it much better... The difficult part is how to shift that mindset imo
Can’t agree more! Besides seeing the possibility of a mindset shift, actually practicing it in daily life is the real gem! More videos will come 😁
FIRST<3
Thank you for watching ❤️
WOW!, I really need your help!
Great work, love the solution for attachment points!
Have you published the source code of this project please ???
Interesting comparison. Which version of Grasshopper were you running?
Hy ,I want to make my own tool that generate designs, but I don't knew how?
Where is this website or APP?! 🤯🤯🤯
Hi I really liked your kitchen layout automation demo.I would like to ask if you could share your dynamo script. Take care, thanks.
Dear PoYen Hsieh, Well that's great! I really liked it and it was extremely inspiring! I would be very interested if there was a teaching material possible that would show you how and in steps you came to the solution. We could learn from the script if you could share it. Thanks
Tutorial please!
Awesome video. Something I can give you as advice. Usually, if you have hard constraints like in your example - having the short furniture together: instead of having this as part of the optimization, just hard code it in. Meaning that you can script the short furniture to always be together, that way you assure that your algorithm is not focused on solving heuristic problems but solving the harder - metaheuristic ones. In this scale, you might skip this because the algorithm will still find a good solution, but as soon as the problem scales, it's always good to reduce heuristic problems and let the algorithm solve the more complex problems.
😠
Hi , love the videos! I’m an architecture students from India and I was planning on doing my thesis to explore with generative design and just needed some guidance regarding. Firstly I use rhino and grasshopper so is this kind of evaluations possible with this software or I need Softwares like dynamo and refinery and secondly where can I learn these Softwares in greater depth
can you share us some of your files ?
Fantastic
This super awesome, can you share a course or stating point to do the study of this super awesome workflow, please, cheers
It's an interesting study
shocked
Hello PoYen, this is incredibly interesting. Would you be able to share the Dynamo script? I would like to learn the logic for moving geometry and creating penalties.
PoYen - Is there anyway to get access to the Dynamo graph? I want to give it a try it and see if I can improve the performance.
Somehow posted this under the wrong video. I was interested in your other video of comparing GH and Dynamo performance
Question: how did you do it go get each kitchen element on a different location? Can you share your Dynamo script?
My God
Great study, what sort of scoring criteria dif you use to test? Edit: found your main video, nevermind. Thanks for sharing!
Parabéns, este estudo mostra o poder computacional que precisamos para agilizar as decisões e melhorar os espaços construídos.