Harnessing the potential of AI for housing renovation

Arti­fi­cial intel­li­gence (AI) is increas­ing­ly used in ren­o­va­tion projects. As such, it reminds us that ren­o­va­tion is an area of inno­va­tion for which we can dis­tin­guish four sep­a­rate areas. 

The first is to clas­si­fy to pri­ori­tise. AI is inte­grat­ed into large-mesh diag­nos­tic tools. For exam­ple, it will make it pos­si­ble to iden­ti­fy – based on ener­gy bills in rela­tion to their size – the 1,000 or 10,000 pub­lic build­ings out of 500,000, which will have the great­est impact per euro invest­ed in terms of reduc­ing CO2 emis­sions or ener­gy consumption. 

The sec­ond area is the cre­ation of opti­mised ren­o­va­tion plans, upstream of a build­ing site: they make it pos­si­ble to improve the lay­out of a flat, the ren­o­va­tion of a build­ing or the renew­al of an urban area. The third area, at the moment of imple­men­ta­tion, is the cen­sus. Com­put­er visu­al­i­sa­tion tech­niques, which rely on images from var­i­ous sources (scans, drones etc.), make it pos­si­ble to take stock of every­thing on a site, but also to deter­mine the per­cent­age of progress or to iden­ti­fy risks.

The fourth area is every­thing that hap­pens when the build­ing exists: pre­dic­tive main­te­nance and intel­li­gent man­age­ment. Based on var­i­ous sen­sors, AI makes it pos­si­ble to antic­i­pate break­downs or to bet­ter under­stand the pref­er­ences of the build­ing’s users. Some smart build­ing sys­tems go fur­ther by allow­ing inter­ac­tion between users (or inhab­i­tants) of the build­ing with con­ver­sa­tion­al agents, or by mak­ing pre­dic­tions about the needs of res­i­dents and thus gen­er­at­ing pro­pos­als for per­son­alised services.

In terms of ren­o­va­tion, it is cur­rent­ly the sec­ond area, design, that is see­ing the most promis­ing advances. 

At the build­ing lev­el, tech­ni­cal stan­dards are mas­tered by pro­fes­sion­als and the con­tri­bu­tion of AI is not yet major. How­ev­er, it will allow for ener­gy per­for­mance, envi­ron­men­tal impact, or cli­mate resilience sim­u­la­tions accord­ing to var­i­ous objec­tives and con­straints. This will con­tribute to opti­mis­ing ren­o­va­tion solu­tions from an eco­nom­ic and envi­ron­men­tal point of view. In a chang­ing envi­ron­ment, where ener­gy and mate­r­i­al prices are evolv­ing rapid­ly as they are now, the use of sim­u­la­tion allows us to reopen the field of pos­si­bil­i­ties and to find ways beyond pro­fes­sion­als’ cur­rent habits. More gen­er­al­ly, automa­tion also allows work to be car­ried out more quickly.

On a neigh­bour­hood scale (more rarely on the scale of an entire city), where things become more com­pli­cat­ed, AI plays a more impor­tant role. Both for ques­tions of resources: the indus­tri­al trades involved are prac­tised by pow­er­ful eco­nom­ic play­ers, often at the fore­front of tech­nol­o­gy; and for ques­tions of needs: it is main­ly at this scale that new vari­ables appear, which were not pre­vi­ous­ly inte­grat­ed and have no obvi­ous “tech­ni­cal solu­tions”, trans­posed into practices.

These issues include the envi­ron­men­tal impact of ren­o­va­tion activ­i­ties and the resilience and adap­ta­tion of infra­struc­tures to cli­mate change. They are added to the more tra­di­tion­al ones that have gov­erned deci­sion-mak­ing until now: cost of the work, com­fort, ener­gy con­sump­tion in the eco­nom­ic sense (pur­chas­ing pow­er of house­holds, costs for man­agers), ener­gy con­sump­tion in the eco­log­i­cal sense (emis­sions). A ren­o­va­tion plan is now a very com­plex matter!

Basi­cal­ly, we are in the busi­ness of deci­sion sup­port. Let’s take an exam­ple. In France and in Europe, many ren­o­va­tion oper­a­tions con­cern neigh­bour­hoods or build­ings belong­ing to the ‘social’ stock, built for the most part in the thir­ty years fol­low­ing the end of the last World War. These are there­fore col­lec­tive deci­sions, involv­ing pub­lic actors and, of course, pub­lic mon­ey. The process­es are cum­ber­some, the deci­sion-mak­ers numer­ous and the stakes high, because the phys­i­cal dete­ri­o­ra­tion of these neigh­bour­hoods goes hand in hand with social and some­times polit­i­cal prob­lems. More­over, they are most often heat loss areas, and the poor qual­i­ty of the build­ings expos­es the inhab­i­tants to ener­gy shocks (increase in price, dis­com­fort). Final­ly, the design of the neigh­bour­hoods and build­ings makes them vul­ner­a­ble to the con­se­quences of cli­mate change, par­tic­u­lar­ly heat waves and flood­ing. There are car parks in these neigh­bour­hoods that are heat pock­ets, and those on the banks of rivers may be sub­ject to a flood risk that was neg­li­gi­ble when they were built.

Ren­o­va­tion is both an urgent and com­plex sub­ject, which rais­es new issues that are not easy for pub­lic deci­sion-mak­ers to master.

Copié !

To sum up, ren­o­va­tion is both an urgent and com­plex sub­ject, which rais­es new issues that are not easy for pub­lic deci­sion-mak­ers to mas­ter. One of the essen­tial chal­lenges is there­fore to accel­er­ate and opti­mise the col­lec­tive deci­sion. This is pre­cise­ly what AI allows.

How can it do so? By allow­ing, dur­ing the ear­ly stages of the dis­cus­sion, us to auto­mat­i­cal­ly gen­er­ate opti­mised rep­re­sen­ta­tions, plans for exam­ple, which are not “the” solu­tion, but which allow the dis­cus­sion to progress. These rep­re­sen­ta­tions are what is known in man­age­ment sci­ence as “bound­ary objects”: they are suf­fi­cient­ly con­crete to con­sti­tute an inter­face between social worlds and actors with dif­fer­ent perspectives.

In a deci­sion-mak­ing process that lasts a dozen months, if we take a worst-case sce­nario, AI could save one or two months. This is sig­nif­i­cant, with­out being a rad­i­cal upheaval. We can also assume that the qual­i­ty of the deci­sion will be bet­ter: it is made in a way that allows every­one to under­stand the rea­son­ing of oth­ers, and on the basis of rep­re­sen­ta­tions that are eas­i­er to dis­cuss than columns of fig­ures. AI, by mak­ing it easy to gen­er­ate pro­pos­als, is a facil­i­ta­tor and opti­miser of human decision-making.

Not in the sense that a com­put­er would speak at the table… but a dis­cus­sion dimen­sion can be inte­grat­ed into the design work that involves AI. The area that inter­ests us in the Ren­o­vAIte R&D pro­gramme on which I am work­ing is what is called Adver­so­r­i­al Resilience Learn­ing (ARL), a con­cept that enables arti­fi­cial neur­al net­works to be mod­elled and trained by putting them in com­pet­i­tive sit­u­a­tions. We take two vir­tu­al enti­ties, which will lit­er­al­ly throw them­selves against each oth­er. One will play the role of defend­er; it will be a mod­el of the urban com­plex based on BIM (Build­ing Infor­ma­tion Mod­el­ing: dig­i­tal mod­els that do not rep­re­sent geo­met­ric shapes, but objects) and oth­er known ele­ments. The oth­er will play the role of the attack­er: this can be any cri­sis (cold, heat, flood­ing, with risks on com­fort, on heat­ing sys­tems, risks of cracks or impreg­na­tion, bud­getary con­straints, etc.). After sev­er­al hours, this com­pe­ti­tion will allow each of the enti­ties to devel­op the best strate­gies and come up with a set of opti­mised urban renew­al plan pro­pos­als for decision-makers. 

It is already work­ing: this con­cept has been used in Ger­many in the design of smart grids, to iden­ti­fy mal­func­tions on the net­work and gen­er­ate the appro­pri­ate auto­mat­ic responses.

Interview by Richard Robert