The past 18 months have produced a parade of prototype ground robots aimed squarely at the problem brigades complain about most: getting supplies, energy, and specialist capability to dispersed units without exposing Soldiers. The machines are no longer laboratory curiosities. They are being driven and pushed in realistic experiments from Fort Irwin to Europe, and companies large and small are pitching modular platforms intended to do what humans do today — carry, clear, and sense — but at lower risk and at scale.
Concrete prototypes fall into three operational families that matter for brigade support. First, the logistics and energy-transport vehicles: purpose-built autonomous tactical transports designed to move multi-class supplies between brigade support areas and dispersed firing units. Overland AI’s ULTRA demonstrations in 2025 showed a platform meant to operate off-road, beyond line of sight, and in degraded navigation conditions while carrying roughly a thousand pounds of payload — the sort of capability brigades need to reduce exposed resupply convoys and to sustain distributed operations. Those demonstrations were conducted as Soldier-led experiments during multinational exercises.
Second, the smaller “robotic mule” category exemplified by the Army’s Small Multipurpose Equipment Transport family. SMET variants and foreign equivalents have been trialed in Project Convergence and live exercises to unease a single operator with an eight-wheeled payload carrier or to act as a squad-level logistic assistant carrying ammunition, batteries, or materiel across complex terrain. These systems are already being used experimentally during combined exercises and are conceptually attractive because they reduce the tactical footprint of human cargo handlers while integrating quickly with existing small-unit workflows.
Third, specialized engineer and breaching robots are moving from concept to formal requirements. The Army’s April 2025 market research and sources-sought for the Breaching and Demolition Ground Engineer Robot, BaDGER, makes explicit what combat engineers have long feared doing with Soldiers alone: autonomous or remotely operated systems that detect, reduce, and proof breaches across wire, berms, and minefields will be procured to remove people from the most dangerous tasks. The BaDGER notices frame the capability as a force multiplier for combined-arms breaching in large-scale combat operations.
Around those categories sit two linked trends that determine operational value. The first is modularity and power export. Modern prototype offers are not single-mission curios. Teams such as AM General teamed with Carnegie Robotics and Textron Systems in October 2025 to propose a hybrid-power, drive-by-wire chassis for the Army’s Medium Modular Equipment Transport concept. That proposal explicitly emphasizes exportable power, MOSA-compliant payload integration, and multi-mission modularity — important because brigade support is increasingly about energy as much as ammunition. A forward element that can receive local power generation from a logistics robot changes sustainment calculus.
The second trend is the march of autonomy away from line-of-sight teleoperation toward higher levels of software-led navigation and perception. Companies debuted systems that fuse LiDAR, stereo vision, inertial measurement, and GPS-denied localization to allow operations in contested or degraded electromagnetic environments. Those autonomy stacks are impressive in demos, but their battlefield performance depends on ruggedized sensing, resilient communications, and the ability to operate safely when inputs fail. Exercise-driven experiments have been crucial because they reveal how brittle autonomy can be when dust, smoke, and intentional jamming are present.
Prototype success does not equal immediate operational utility. There are three practical gaps brigade commanders will feel first. One, survivability and attritability. Logistics robots will be high-value targets. If the Army fields expensive, long-lead systems without an attritable model or a deliberate plan for replenishment, commanders will either hoard them behind the brigade support area or lose tempo trying to protect assets that should be expendable. Two, human-machine workflows. Trials show soldiers adapting tactics around robots; that is positive. But doctrine and training lag. Small-unit leaders need playbooks that cover robot failure modes, handoffs, and rules for recovery under fire. Three, sustainment and repair. A ground robot that cannot be fixed with the tools and parts a forward support company carries is a paperweight. Prototypes must be designed with field maintenance and spares economics in mind.
There is also an ethical and legal dimension that cannot be papered over by better sensors. Some quadrupedal prototypes have been fitted with remote weapon turrets and counter-UAS sensors during experiments and exercises. The presence of weaponized or weapon-capable platforms in brigade support mixes complicates command responsibility and the practical separation between combat support and combat systems. The moral calculus of removing soldiers from risk is attractive. But it becomes morally opaque if the machines we place forward have the capacity to apply lethal force without clearly auditable human decision chains. Maintaining human-in-the-loop control for engagements remains a necessary restraint if we care about accountability and international norms.
From these prototypes a set of design prescriptions follows. First, favor modular, MOSA-friendly architectures so brigades can choose payloads that meet mission-specific sustainment, sensing, or engineer needs rather than buying bespoke single-purpose vehicles. Second, design for attritability and low-cost recoverability: replaceable sensor suites, swappable powerpacks, and commonality of chassis across classes reduce logistic strain and allow commanders to accept losses without cascading mission failure. Third, build autonomy suites with constrained “fail to safe” behaviours that prioritize retrieval and denial over independent action when localization or comms are lost. Fourth, pair procurement with doctrine development and a robust, iterative experimentation pipeline so TTPs evolve as fast as the hardware. Fifth, insist on human-centred maintenance: every system fielded to a brigade must be maintainable by the organic support chain within the time and resource envelope of deployed sustainment units.
The final point is institutional. Prototypes such as ULTRA, SMET families, and the BaDGER notices show an institutional willingness to experiment and to buy capability-driven prototypes rather than single-minded platform programs. That is the right posture. But procurement must avoid two errors: overpromising on autonomy maturity and under-resourcing the non-glamorous parts of robotics integration, namely spares, training, and cyber-electronic protection. If brigades are to be supported by robots rather than enchanted by them, the services must fund the plumbing behind the demos. Otherwise the next generation of so-called brigade support robots will be parade pieces, not logistics enablers.
We are at an inflection point where prototypes are credible enough to change tactics, but doctrine and industrial practice must catch up. The sensible path is iterative adoption: field small, modular fleets; embed repair specialists in brigades; update tactics continuously based on live experiments; and codify human control and accountability for any weapon-capable systems. If we do that, ground robots will deliver on their promise to reduce risk and increase operational reach. If we do not, we will have advanced the technology while leaving the human institutions that must use it behind.