Military ground robots : a technological shift ?

What are the chal­lenges of mili­ta­ry land robo­tics today ?

David Filliat. Whe­ther mili­ta­ry or civi­lian, robo­tics covers dif­ferent tech­ni­cal domains. There are sen­so­ry tech­no­lo­gies or mobi­li­ty tech­no­lo­gies, from a mecha­ni­cal or arti­fi­cial intel­li­gence point of view, etc. Impres­sive pro­gress has been made in recent years : Bos­ton Dyna­mics’ robots, for example, are very suc­cess­ful in terms of phy­si­cal mobi­li­ty and are ful­ly ope­ra­tio­nal. But a lot remains to be done if we consi­der the constraints of the bat­tle­field or mili­ta­ry ope­ra­tions in a broa­der sense – both in terms of dis­cre­tion and robust­ness, we are only at the begin­ning of this story.

The grea­test chal­lenges today relate to auto­no­mous move­ment. Unk­nown and com­plex ter­rain has lit­tle in com­mon with a road where, broad­ly spea­king, there are only pedes­trians and cars. In a mili­ta­ry context the ter­rain is more rug­ged and varied. The need to be able to navi­gate using GPS coor­di­nates remains the same, but when it comes to the details of plan­ning the route the sligh­test branch can be an obs­tacle that must be detec­ted, eva­lua­ted, and mana­ged – either by soft­ware or mecha­ni­cal­ly. So, there are addi­tio­nal, higher-level challenges.

Rather than huma­noid (or dog-like) robots, we should ima­gine whee­led or tra­cked robots and the robo­ti­sa­tion of equipment.

Howe­ver, desi­gners ima­gine com­plete sys­tems, with each method pro­vi­ding solu­tions to cir­cumvent cer­tain pro­blems. In robo­tics, for example, aerial sys­tems can be sim­pler than ter­res­trial sys­tems. Howe­ver also brings a form of vul­ne­ra­bi­li­ty, unless one can fly very high, far from ope­ra­tions control, which involves tech­ni­cal choices. If we consi­der the on-board capa­ci­ty, ground vehicles (ver­sus wal­king or flying) allow us to car­ry more ener­gy and include more sen­sors. Each sys­tem has its own advan­tages and disad­van­tages ; at the moment, there is no single ‘best’ solu­tion. We think more in terms of the dif­ferent applications.

Is full auto­no­my out of reach today ?

On the bat­tle­field, cer­tain­ly. It is not in the coming years that we will see ful­ly auto­no­mous robots car­rying out com­plex mis­sions. Robots will pro­ba­bly appear along­side humans. Rather than huma­noid (or dog-like) robots, we should ima­gine whee­led or tra­cked robots and the robo­ti­sa­tion of equipment.

For example, we will see dri­ving aids in the form of convoys, as is alrea­dy being expe­ri­men­ted with in the civil sec­tor with convoys of semi-trai­lers, or in the form of self-dri­ving vehicles that take over from humans on open ter­rain. This lat­ter could allow the dri­ver to leave controls and rest for an hour.

The main objec­tive in all these tech­ni­cal efforts is to save human lives. But this goes hand in hand with the search for other, dif­ferent, more effi­cient ways of acting on the ene­my. Robo­tics also comes into play in a context where new threats are appea­ring, which are them­selves dehu­ma­ni­sed, such as swarms of drones. This is the­re­fore a ques­tion of pro­tec­ting one­self, pro­tec­ting dri­vers and ground sol­diers, by mecha­ni­sing and auto­ma­ting cer­tain tasks.

Robo­tics is only a new stage in what is alrea­dy a long his­to­ry. During the Second World War, the Ger­mans desi­gned the V1 to limit the loss of pilots, which are rare and high­ly qua­li­fied per­son­nel. Today’s armies rea­son in the same way with all troops, because in dis­tant conflicts that are not ful­ly legi­ti­mate in the eyes of public opi­nion, a single human life lost has a great impact.

Howe­ver, we must not pre­tend that we are not wit­nes­sing changes. Fif­teen years ago, the appli­ca­tions of robo­tics were still logis­tics, obser­va­tion, and sup­port. Today there is a switch towards to letha­li­ty. This effort is due in part to the per­cep­tion that the threats faced have evol­ved : in the short term, the per­cei­ved dan­ger comes from swarms of small drones. But large coun­tries can deve­lop lar­ger, steal­thier drones : the deve­lop­ment of auto­no­mous wea­pons sys­tems is a res­ponse to this threat, in an inevi­table tech­no­lo­gi­cal race.

What are the deter­mi­nants of this tech­no­lo­gi­cal race ?

An essen­tial and new ele­ment, for me, is that we are tal­king about dual tech­no­lo­gies, which are not being deve­lo­ped sole­ly for the mili­ta­ry. This is very dif­ferent from what hap­pe­ned in the nuclear or aero­nau­ti­cal sec­tors, not to men­tion the many tech­no­lo­gies deve­lo­ped within the fra­me­work of the Ame­ri­can DARPA that were then dis­se­mi­na­ted to the civi­lian world (think of the Inter­net, GPS, etc.). Today, it can even be consi­de­red that the civi­lian world (indus­tria­lists, major soft­ware players) is a mel­ting pot for tech­no­lo­gies that could revo­lu­tio­nise this type of equip­ment. The mili­ta­ry have unders­tood this and are loo­king for syner­gies with the civi­lian world.

The mili­ta­ry are loo­king for syner­gies with the civi­lian world.

This is the case when it comes to the abi­li­ty to ana­lyse sur­roun­dings, where algo­rithms must be capable of pro­ces­sing huge volumes of data. This is where the civi­lian world may have an advan­tage as the num­ber of sys­tems deployed that can acquire data is much grea­ter. Auto­no­mous vehicles that car manu­fac­tu­rers and equip­ment sup­pliers are cur­rent­ly wor­king on must deal with deci­sion-making, tra­jec­to­ry plan­ning and per­cep­tion pro­blems, which are all func­tio­na­li­ties that are also essen­tial for mili­ta­ry use. Even if, as we have seen, the chal­lenges are grea­ter in a mili­ta­ry context, and the data more dif­fi­cult to col­lect, the same tech­no­lo­gi­cal buil­ding blocks are being developed.