Urban AI | Experiments in river Llobregat is a study in the use of machine learning algorithms and Artificial intelligence as a tool to generate ideas in urban space. To explore this tool we are using our design studio project that is focused on renaturalizing the river LLobregat in Barcelona. The site is located in the lower Delta area and is a predominantly agricultural land. We are looking to adapt this low lying area to mitigate floods using different strategies from river meandering to wetland creation.

llobregat river, renaturing, river delta, sinay coskun, laura guimaraes, mario jose gonzalez, hebah qatanany, aishath nadh ha naseer

Location of the River & Area of Intervention

Using ‘Ceasar’ as a tool to simulate how and where the river would potentially flow and where the sediment would transfer, we were able to generate a series of images showing the simulation of the river over different time spans.

llobregat river, renaturing, river delta, sinay coskun, laura guimaraes, mario jose gonzalez, hebah qatanany, aishath nadh ha naseer, caesar, simulations

Outputs from Caesar Simulations

At first, we started with a few non-guided spatial experiments where we used the existing site as the content without the new meander. We ran a few jobs using a dense urban texture as a style and tried different iterations to arrive at the output. Next, we tested the same content with images of wetlands to try to visualize how and where the alternative geographies we are proposing would start to appear.

llobregat river, renaturing, river delta, sinay coskun, laura guimaraes, mario jose gonzalez, hebah qatanany, aishath nadh ha naseer, urban AI, artificial experimants, style transfer

Non-guided Spatial Experiments

Going one step further, we wanted see how the algorithm would steer us if we introduce our initial proposals at a larger scale. The final result reveals our proposal that puts forward the wetlands and new channels simulated by Caesar. Since one of the main objects of the project is combining agricultural production with river and the entire city via river; we tried to see more and more ‘green’ in a combination with the existing urban patterns around the river. And once we concretize our work using z brush, we had a clear idea on how the river is included in the whole design composition and what kind of land subdivision or parceling can be implemented in the area of intervention.

llobregat river, renaturing, river delta, sinay coskun, laura guimaraes, mario jose gonzalez, hebah qatanany, aishath nadh ha naseer, urban AI, artificial intelligence, style transfer

Guided Spatial Experiments

On a smaller scale, we focused on the pockets created in between our meanders on different time spans, in order to visually test out our proposals on site over time. Therefore, we made several experiments applying aquaculture fields as a style, for example rice fields with various colours and textures, which gave us a simulation of what the site could look like with our project.

llobregat river, renaturing, river delta, sinay coskun, laura guimaraes, mario jose gonzalez, hebah qatanany, aishath nadh ha naseer, urban Aı, artificial intelligence, style transfer

Through these experiments, we were able to understand the methodology behind this algorithm and its capability as a powerful tool to quickly generate numerous possible urban scenarios that can be used for further research and investigation.

llobregat river, renaturing, river delta, sinay coskun, laura guimaraes, mario jose gonzalez, hebah qatanany, aishath nadh ha naseer, urban AI, artificial intelligence, style transfer

All experiments made throughout the workshop

Urban AI | Experiment in river Llobregat is a project of IAAC, Institute for Advanced Architecture of Catalonia developed at Master in City & Technology in 2020/21 by student: Aishath Nadha Ha Naseer, Hebah Qatanany, Laura Guimaraes, Mario Jose  Gonzalez, Sinay Coskun and faculty: Prof. Sandra Manninger & Dr. Matias del Campo