solution-oriented collaborative computational planning for smoother decision-making

WUF10: Connecting Culture & Innovation

Agent-based cultural behaviour simulation might contribute to consider more realistically social phenomenon to allow the conception of better collaborative urban planning tools.

As it was stated in WUF10 concept paper and discussed in many events during the forum, the major driver of recent urbanization is migration, which makes cities more and more diverse and heterogeneous. Multicultural metropolises are thus of higher complexity in terms of urban management. Trying to understand the influence of culture and cultural diversity in cities and analyse it in a holistic matter is one of the challenges addressed in WUF10.

Cutting-edge innovations in terms of spatial simulation can help us to understand more precisely those kinds of very complex and intricate phenomena. The approach of agent-based modelling, made possible through computer programming, first comes to my mind when it comes to simulating cultural behavior.

If you’re not familiar to agent-based modelling, here is what it basically does.

Sometimes we might not perfectly understand how the system as a whole behave, what are the key variables and dependencies between them, but we may have some insights about how each object in the system behave individually. Therefore, we can start building the model from the bottom-up by identifying those objects (or agents) and defining their typical behavior.

Behavior is one of the key components of culture, even though it is often not considered and studied, due to the fact that it is very hard to rationalize and predict, and that it is not what we can call a politically correct matter.

Agent-based modelling approach, which is more and more adopted in environmental and social studies, can also benefit to urban planning as a key approach to consider cultural behaviors. Two types of usage of that approach that first comes to my mind: (1) agent in environment, and (2) interaction of agents.

(1) We can put the agents in an environment which has its own dynamics. The global behavior of the system then emerges out of many (tens, thousands, even millions) concurrent individual behaviors.

For example, let’s consider the example of assessing the impact of livestock grazing in the context of rarefaction of natural resources (fodder and water), which is exactly the case of the Sahel region in Western Africa. Livestock grazing by nomadic population is a traditional activity, but climate change and migrations put the natural resources at stress. Grazing activity and routes as well as resources regeneration are extremely complex, and it would be almost impossible to quantify them holistically for a whole region.

But in the agent-based modelling approach, the characterization of individualities can help us to grasp the whole system. We basically create two agent populations:

  • A population livestock grazing agents (pastor with his/her herd), which cultural behavior can be programmed as followed: the pastor goes from his base camp in the morning >> makes his stock drink at some dedicated water source points >> the stock eats fodder >> when the stock gets thirsty they get to the next water source point in the way >> the pastor and stock comes back to camp. Of course, this behavior shall be documented based on actual surveys and interviews and some additional randomness shall be added to the mix.  ​
  • The fodder resource agent (vegetation) as a discrete space (a grid) with a value in each cell corresponding to the percentage of the fodder in the area: when the fodder is eaten by the herd, the vegetation value decrease, but it regenerates with time based on seasonal rainfall. Overgrazing and trampling would be represented by a negative value of the cell, corresponding to a soil erosion (an irreversible impact).

Using this kind of simulation, we would be able to assess relatively realistically the impact of pastoralism over a certain local environment.

Now, in the perspective of solution-oriented spatial planning, we could add to this simulation the possibility of adding new water infrastructure such as livestock watering spots for example. Policymakers and experts would interact with the model to simulate the future impact of the construction of new livestock watering system.

(2) The second example of application of agent-based cultural behavior programming to sustainable urban planning is the possibility of letting the agents to interact with each other in a system, for example a transportation network. Indeed, agent-based modelling is increasingly used in mobility analysis and planning.

This type of simulation could be used, in our perspective of real-time collaborative planning, to simulate on the fly the impact of our urban mobility planning policies (closing roads, reducing speed etc.) on the system of agents.

The example in the video is using Jung’s routing algorithm which gives a cost to all the possible roads that agents are able to take: similarly to human agents using a GPS application, agents in the simulation will avoid sections with speed limitation, sections congested by other fellow agents, and of course no entry roads (even though some agents might not respect the rule…just like in real life).

The construction of robust and realistic agent-based mobility models can help urban planners to forecast and mitigate the future impact of the increasing demand for urban logistics in very dense old fabrics (quartier Saint-Michel in Bordeaux, France in the video).

February 12, 2020

Photo 1: en Haut!; Photo 2: détoursdumonde