Pro­fes­sor Philippe Bon­ni­fait is also Direc­tor of the Heudi­asyc lab­o­ra­to­ry, a UMR (UTC/CNRS) spe­cial­iz­ing in com­put­er sci­ence as well as robot­ics applied to drones and smart vehi­cles. This exper­tise is wide­ly rec­og­nized, as the lab­o­ra­to­ry was invit­ed to par­tic­i­pate in the PEPR “Robot­ics” pro­gramme, sched­uled to run for six and a half years, on the PERSEO theme (€3.3 mil­lion), which brings togeth­er 15 lab­o­ra­to­ries, includ­ing UTC’s Heudi­asyc Laboratory.

This is the lat­est PEPR approved, in June 2025, as part of the “France 2030” incen­tive. Why “Robot­ics Accel­er­a­tion”? “There are two types of PEPRs. There are those that are sci­en­tif­i­cal­ly ambi­tious but are more explorato­ry in the sense that, in cer­tain fields, we observe that break­throughs could emerge but with­out an asso­ci­at­ed inno­va­tion path­way vis­i­ble at this stage. This is the case, for exam­ple, with the “Organ­ic Robot­ics” PEPR ‚” which brings togeth­er robot­ics engi­neers and social sci­ence spe­cial­ists. And there are those PEPRs that already have an estab­lished inno­va­tion path­way and are there­fore con­sid­ered strate­gic for the coun­try, with the goal is to “accel­er­ate” the inno­va­tion move­ment. This is the case with the “Robot­ics” PEPR. Robot­ics is a mul­ti­dis­ci­pli­nary sci­ence that draws upon, inter alia, mechan­ics, com­put­er sci­ence and mecha­tron­ics, which is evolv­ing rapid­ly in step with tech­no­log­i­cal inno­va­tions. “It is a trans­for­ma­tive pro­gramme that will define what French robot­ics will look like in five years,” explains Philippe Bon­ni­fait, who served on the “Robot­ics” PEPR’s found­ing com­mit­tee and is now a mem­ber of its steer­ing committee.

A PEPR that received near­ly 130 calls for expres­sions of inter­est (CEI). “No AMI was select­ed as such. We analysed all the pro­pos­als and then syn­the­sized them. We then chose the CEIs with the most inno­v­a­tive pro­pos­als and offered them a rather dis­rup­tive orga­ni­za­tion­al mod­el. The goal was to pre­vent each lab­o­ra­to­ry from work­ing blind­fold on its own project but rather cre­at­ing an ecosys­tem that brings togeth­er everyone’s inno­v­a­tive ideas to pre­pare for the trans­for­ma­tion of French robot­ics in light of inter­na­tion­al chal­lenges and devel­op­ments. Hence, we defined five research areas, includ­ing MINIRO on minia­ture robot­ics, robot manip­u­la­tion in indus­tri­al set­tings, the embod­i­ment of AI in robot­ics and PERSEO, led by Joelle El Hage at Heudi­asyc, ded­i­cat­ed to “safe coop­er­a­tive per­cep­tion in open and evolv­ing envi­ron­ments for robot­ic auton­o­my” ‚” he explains. Indeed, with the rise of AI, the objec­tives of the “Robot­ics” PEPR focus on devel­op­ing capa­bil­i­ties of robots in the areas of per­cep­tion, grasp­ing, con­trol­ling, auton­o­my and, final­ly, resource effi­cien­cy. These objec­tives res­onate with Philippe Bonnifait’s work. “My per­son­al research focus­es pri­mar­i­ly on issues of per­cep­tion, local­iza­tion and map­ping for autonomous vehi­cles,” he says.

What are some of PERSEO’s objec­tives? “It involves study­ing the issue of per­cep­tion for robots, cur­rent­ly far from resolved. With­in the frame­work of PERSEO, we are inter­est­ed in main­ly gener­ic fields of per­cep­tion that could apply to autonomous robots oper­at­ing in uncon­trolled envi­ron­ments, as opposed to those used in ware­hous­es or fac­to­ries, for exam­ple. These are essen­tial­ly mobile robots, often equipped with wheels, such as cars, trucks, bus­es, or even agri­cul­tur­al machin­ery but they can also be aer­i­al or aquat­ic robots, whether under­wa­ter or sur­face, to be used in the main­te­nance of off­shore wind tur­bines, for exam­ple,” empha­sizes Philippe Bonnifait.

Robot auton­o­my requires that they be able to per­ceive and under­stand their envi­ron­ment. “PERSEO must there­fore address two main chal­lenges: envi­ron­men­tal under­stand­ing and data fusion. To devel­op per­cep­tion sys­tems, it is indeed nec­es­sary to com­bine sev­er­al types of sen­sors. We will also address two more trans­verse areas. These involve defin­ing meth­ods for test­ing algo­rithms and per­cep­tion sys­tems using data from spe­cif­ic envi­ron­ments or, as is increas­ing­ly the case, using sim­u­la­tion envi­ron­ments with dig­i­tal twins. Per­cep­tion is one of the key pil­lars of auton­o­my, since the robot must be able to form a rep­re­sen­ta­tion of the world and under­stand what is hap­pen­ing around it. And if the robot oper­ates in a dynam­ic world—in the sense that oth­er actors share the same space, it must also be able to pre­dict each other’s move­ments,” he explains.

No few­er than ten the­sis top­ics will be launched as part of PERSEO. The 15 lab­o­ra­to­ries will work in pairs on com­mon top­ics to fos­ter nation­al syn­er­gy and cre­ate a trans­for­ma­tive effect. “For exam­ple, a the­sis on per­cep­tion algo­rithms will be co-super­vised by the École des Mines and ENSTA. Anoth­er, co-super­vised by UTC-Com­pieg­ne and Ecole Cen­trale Nantes, aims to answer the fol­low­ing ques­tion: will the data pro­vid­ed by per­cep­tion allow the robot to accu­rate­ly locate itself and cre­ate a geo­graph­i­cal rep­re­sen­ta­tion of its envi­ron­ment? Oth­ers are under dis­cus­sion with INSA Rouen, the Uni­ver­si­ty of Lille and the Uni­ver­si­ty of Toulouse,” con­cludes Philippe Bonnifait.

MSD