Résumé
Agroforestry is recognized as a way of developing a sustainable and resilient agriculture that can fight against climate change. However, the number of species combinations, spatial configurations and management options for trees and crops is vast. These choices must be adapted to the pedoclimatic and socio-economic contexts and to the objectives of the farmer, who therefore needs support in designing his system. Participative design workshops are a good way to integrate the knowledge of several experts in order to design such complex systems. The design of agroforestry systems should take into account both spatial aspects (e.g. spacing of trees within the lines and between lines, tree line orientation, tree-crop distance, species spatial patterns) and temporal aspects (e.g. crop rotations, tree thinning and pruning, tree planting in the case of successional agroforestry). Furthermore, the interactions between trees and crops evolve as the trees grow. However, agroforestry design workshops generally emphasize the spatial aspect only, through the use of static tokens to represent the different species when designing the spatial configuration of the system.New technologies could remove this limitation. In particular, augmented reality (AR, the superimposition of digital objects on real-world images) makes it possible to visualize dynamic representations of trees and crops, and also their interactions, while at the same time retaining the possibility to physically interact with the system being designed (i.e. move trees, add or remove species etc.).Our objective is to propose an ergonomic digital solution capable of assisting a group of agroforestry experts to design an agroforestry system. It should enable the group of experts to understand the dynamics and functioning of the system over time and to assess its landscape impact. Several digital technologies can be used to implement AR: marker-based AR (a specific marker, or pattern, existing in real life is used by the app to detect where the digital object should be displayed on the screen, geolocation-based AR (the app uses the telephone sensors, such as GPS and accelerometer, to compute where the digital object should be displayed) or markerless AR (the app automatically detects real-life objects, such as walls or the ground, through the camera to compute where the digital object should be displayed).In this study, we investigated the use of marker-based AR to develop a digital tool able to assist in agroforestry systems design workshops. Since the workshops gathers people from different backgrounds, we looked for digital solutions that do not require specific hardware (apart from a smartphone), and ideally that would not even require installation of specific apps, so that all users could use their own smartphones right out of the pocket. As a result, we turned to web-based AR tools, which use the phone's web browser to display the digital objects on the screen. Our first choice was to use AR.js, a lightweight library for Augmented Reality on the Web that is open-source and free of charge. We designed a first prototype that is able to detect around 20 different markers, although for now the prototype displays only two digital objects (a 3D model of a tree and blocks of wheat-looking crops). The limitations and constraints of this tool in the context of agroforestry design workshops are being evaluated.Current work concerns:- the number of tree species available in the application : 3D digital models for walnut, poplar, wild cherry and other species used in agroforestry systems have been developed and are in process of being integrated in the prototype;- the representation of trees at various growth stages, to see the dynamics of trees as the system ages;- the possibility to have different tree generations, to be able to visualize successional agroforestry systems.Future work will focus on adding features to this prototype such as:- analyzing the positions of objects to extract automatically the patterns and model the agroforestry system;- designing interfaces related to cultural practices, such as thinning or pruning;- representing tree-crop interactions. In a first step, we will focus on tree shade (light competition), although the final objective will be to be able to represent also below-ground competitions (water, nitrogen), or other variables of interest for the design of agroforestry systems (e.g. predicted crop yield).Such a tool could be useful not only in agroforestry design workshops, but also as a pedagogical tool, and its use will be tested with farmers, agricultural extensionists and students in agronomical schools, to see how augmented reality contributes to the understanding of the complexity of interactions in agroforestry systems.