2022-04-roadia

Safer roads and smoot­her traf­fic flow

How ROA­DIA is trans­forming traf­fic manage­ment using arti­fi­ci­al intelligence

The so-cal­­led “Visi­on­Ze­ro” is the visi­on of zero traf­fic fata­li­ties on Euro­pean roads. An ambi­tious goal that the EU Com­mis­si­on is aiming to reach by 2050. The capi­tal regi­on has also set its­elf such tar­gets. At the moment, that is still a long way off. In Febru­ary 2022, for exam­p­le, the Ber­­lin-Bran­­den­­burg Agen­cy for Sta­tis­tics regis­tered a 44.6 per cent increase in traf­fic acci­dents on Berlin’s roads com­pared to the pre­vious month. And Bran­den­burg also saw 16.4 per cent more peo­p­le being invol­ved in acci­dents in Febru­ary than was the case in Janu­ary 2022.1 Even though the sta­tis­tics may fluc­tua­te, we are still a long way from the mobi­li­ty tur­n­around and Smart­Ci­ties who­se infra­struc­tu­re would be gui­ding us safe­ly through traf­fic on the streets.

The Media­Tech Hub Acce­le­ra­tor start-up ROA­DIA would like to chan­ge this and is working on intel­li­gent sen­sors to pro­vi­de more safe­ty in road traf­fic. And not only that. Using its hard­ware and arti­fi­ci­al intel­li­gence (AI), the traf­fic con­trol cent­re of the future can be deve­lo­ped tog­e­ther with muni­ci­pa­li­ties and com­mu­ni­ties. In order to achie­ve this goal, ROA­DIA is working on two curr­ent­ly inde­pen­dent solu­ti­ons that will be mer­ged at some time in the future into an intel­li­gent all-in-one sen­sor. One solu­ti­on sees a traf­fic moni­to­ring sen­sor to replace manu­al traf­fic coun­ting, and the other one is a speed mea­su­re­ment sys­tem which is curr­ent­ly being deve­lo­ped and about to be certified.

The idea came to one of ROADIA’s co-foun­­ders Tobi­as Fischer when he was again nar­row­ly over­ta­ken by a car whilst out on his bicy­cle. Traf­fic sta­tis­tics do not record such “near-acci­­dents”. But safer mobi­li­ty could be ensu­red if one was able to iden­ti­fy and eli­mi­na­te dan­ger spots for dif­fe­rent road users ear­ly on.

AI-based sen­sors ins­tead of manu­al traf­fic counting

In fact, the­re is not­hing new about data coll­ec­tion in traf­fic, but it’s deter­mi­ned by dif­fe­rent actors and in a varie­ty of ways. In tra­di­tio­nal traf­fic coun­ting, for exam­p­le, peo­p­le sit at road junc­tions with clip­boards and pens and record the num­ber and type of pas­sing vehic­les. Today’s traf­­fic-light cir­cuits, for exam­p­le, work via induc­ti­ve loop sen­sors in the ground that have to be instal­led the­re in expen­si­ve con­s­truc­tion mea­su­res and requi­re a high level of main­ten­an­ce. The asphalt is torn open, which is some­thing that can lead to dis­rup­ti­ons to both the traf­fic flow as well as the cables and pipes instal­led the­re. ROA­DIA finds a reme­dy for this by using AI-based sen­sors. They detect the various vehic­le cate­go­ries and thus save muni­ci­pa­li­ties valuable resources.

Speed mea­su­re­ment sys­tems are ano­ther important fac­tor for impro­ving road safe­ty. Cur­bing exces­si­ve speeds also redu­ces the dan­ger on the roads. The mea­su­re­ments are recor­ded by means of RADAR, LIDAR, lasers and light bar­riers. Dif­fe­rent har­d­­wa­re-based com­pon­ents need to work tog­e­ther to achie­ve a rea­ding. A sen­sor gene­ra­tes the mea­su­re­ment, a came­ra docu­ments it. Howe­ver, the came­ra can­not record exact­ly what the sen­sor is actual­ly doing. It only shows the frame within which a sen­sor is mea­su­ring vehic­les or are­as. This can­not be cle­ar­ly moni­to­red if other vehic­le parts are influen­cing the mea­su­re­ment or if the came­ra ang­le has shifted due, for exam­p­le, to a crash. ROA­DIA the­r­e­fo­re reli­es on ima­ging mea­su­re­ment by using its own tech­no­lo­gy. What the came­ra opti­cal­ly detects and dis­plays cor­re­sponds exact­ly with what is actual­ly being mea­su­red. This does not requi­re ano­ther exter­nal sen­sor sin­ce the came­ra mea­su­res the vehic­le pre­cis­e­ly. As a part of the raw data, the­se so-cal­­led mea­su­ring points are later com­pre­hen­si­ble and trans­pa­rent for all of tho­se involved.

Pre­cise traf­fic mea­su­re­ment saving cos­ts and time

The sen­sors that ROA­DIA is deve­lo­ping here in Pots­dam are AI-based. The con­cept is simi­lar to that for self-dri­­ving cars, except for the fact that they are sta­tio­na­ry rather than a moving sys­tem. The traf­­fic-moni­­to­ring sen­sors detect an object moving in space and can pro­cess it. For exam­p­le, a major pro­ject under­ta­ken by ROA­DIA on behalf of the sta­te of Schles­­wig-Hol­stein last year saw the sen­sors being used simul­ta­neous­ly at 140 mea­su­ring points to record traf­fic data. The resul­ting relia­ble data are important for indi­ca­ting to the cities and muni­ci­pa­li­ties the ans­wers to the fol­lo­wing ques­ti­ons: whe­re is the traf­fic beco­ming con­ge­sted? When do roads need main­ten­an­ce? Whe­re should roads be widen­ed becau­se the traf­fic has increased?

“We auto­ma­te this pro­cess,” co-foun­­der Mar­kus Hant­sch­mann explains. “We make sen­sors intel­li­gent and train our own neu­ral net­works. To do this, we crea­te our own soft­ware that can cal­cu­la­te on the hard­ware its­elf and gene­ra­te infor­ma­ti­on that can also be fur­ther pro­ces­sed in real time if necessary.”

This goes bey­ond pure digi­tal deve­lo­p­ment. Data coll­ec­tion is ful­ly auto­ma­ted and more cost-effi­ci­ent. By coll­ec­ting the data local­ly on the sen­sors, one also save having to go the round­about rou­te via a cloud which works rela­tively slow­ly. “We can also mea­su­re what was pre­vious­ly not mea­sura­ble. Distances such as the late­ral distances when over­ta­king, for exam­p­le,” Hant­sch­mann says. This had pre­vious­ly only been done in the case of black spots. Distances are then drawn by hand on the ground and checks are made to see whe­ther the set mar­kings have been dri­ven over.

Detec­ting and avo­i­ding acci­dent black spots

Even though ROADIA’s tech­no­lo­gy can be used world­wi­de: every coun­try, in fact every fede­ral sta­te, has dif­fe­rent gui­de­lines accor­ding to which traf­fic sur­veys ope­ra­te. The Pots­­dam-based start-up tracks what has been defi­ned before­hand. Typi­cal clas­ses, i.e. clas­si­fi­ca­ti­ons of vehic­les that are often tra­cked, include the 5 plus 1 clas­si­fi­ca­ti­on which covers motor­bikes, mopeds, hea­vy goods vehic­les and buses as well as cars. Moreo­ver, ROA­DIA can distin­gu­ish bicy­cles from car­go bikes. It needs the same distance and start­ing posi­ti­on so that the under­ly­ing arti­fi­ci­al intel­li­gence can pro­cess the data. Tra­jec­to­ry cal­cu­la­ti­on can be used to show whe­re which vehic­le is loca­ted at which par­ti­cu­lar time, which lanes and rou­tes are being used more fre­quent­ly and, thus, what the traf­fic situa­ti­on is like at the respec­ti­ve mea­su­ring point. For exam­p­le, one can see whe­re things are beco­ming pre­ca­rious if two lanes keep over­lap­ping. Typi­cal acci­dent black spots in cities include schools and kin­der­gar­tens as well as reti­re­ment homes and other sec­tions of roads whe­re traf­fic speeds chan­ge. The intel­li­gent algo­rith­ms can detect vehic­le cate­go­ries with grea­ter accu­ra­cy than hither­to and be sup­ple­men­ted by indi­vi­du­al­ly defi­nable sub-cate­­go­ries. “In the future, it will also be pos­si­ble to mea­su­re bicy­cle traf­fic and inte­gra­te this into traf­fic light con­trol. Or set up traf­fic lights in such a way that they react to the onco­ming blue lights of emer­gen­cy vehic­les and direct­ly switch the traf­fic lights for the cross traf­fic to red. The pos­si­bi­li­ties are the­re, the imple­men­ta­ti­on must, of cour­se, be tes­ted and pre­cis­e­ly defi­ned,” Hant­sch­mann says.

The traf­fic con­trol cent­re of tomor­row is digital

In the long term, ROA­DIA would like to fur­ther enhan­ce the infra­struc­tu­re in Ger­ma­ny and Euro­pe. The team is curr­ent­ly working on a pro­­ject-by-pro­­ject basis. While speed and distance mea­su­re­ments are inte­res­t­ing for poli­ce and regu­la­to­ry aut­ho­ri­ties, impro­ve­ments to the infra­struc­tu­re are a sub­ject for the pri­va­te sec­tor. Ano­ther exci­ting deve­lo­p­ment are the so-cal­­led real lab pro­jects which are being used to explo­re (digi­tal) mobi­li­ty in various cities wit­hout any regu­la­ti­ons. Enhan­cing the infra­struc­tu­re and impro­ving the inter­ac­tion of all road users not only ensu­res that the­re is a smooth and safe traf­fic flow, but is also of bene­fit for the envi­ron­ment. For exam­p­le, stop-and-go traf­fic is the main cul­prit for CO2 emis­si­ons on our roads.

The ROA­DIA team keeps working meti­cu­lous­ly on the mea­su­ring tech­no­lo­gy and test­ing the sen­sors on Ger­man roads. Moreo­ver, fol­lo­wing their suc­cessful par­ti­ci­pa­ti­on in the Media­Tech Hub Acce­le­ra­tor, they are now based in Pots­dam with offices in the Media­Tech Hub Space whe­re they regu­lar­ly exch­an­ge prac­ti­ces and know­ledge with other start-ups. Legal advice or docu­ments for hiring staff are quick­ly pas­sed from hand to hand here. The chal­lenges for start-up teams are sel­dom to be found in deve­lo­p­ment or mar­ket ent­ry - the hurd­les are usual­ly else­whe­re. Accor­ding to Hant­sch­mann, ROA­DIA was sur­pri­sed that the pro­blem was not in fin­ding staff, but rather in inte­gra­ting the new employees in the metro­po­li­tan regi­on - they always try to be the­re to pro­vi­de sup­port in all mat­ters from regis­tra­ti­on for­ma­li­ties through to fin­ding accommodation.

After all, ROA­DIA is gro­wing and now has employees coming from eight dif­fe­rent count­ries, con­ti­nu­al­ly deve­lo­ping mar­ke­ta­ble solu­ti­ons and see­king to crea­te part­ner­ships with muni­ci­pa­li­ties and other mar­ket players.

About MTH Blog

The media technologies of the future are already being used today – not only in the entertainment sector, but also in a wide variety of industries. Christine Lentz meets up with tech enthusiasts, established companies and researchers for our monthly MediaTech Hub Potsdam blog to tell the stories behind the innovative business models.