XTRACKS-ROBOTICS XTMV-800 UGV

The XTMV-800 Nano MK.II serves as a research tool for academics and engineers,
as much as a cost-effective monitoring or swarming tool
(All images: XTRACKS-ROBOTICS)

Stay on track

Rory Jackson investigates this small, modular tracked vehicle, prized across education and research for its stability, but also poised for first response and attritable defence applications thanks to its low cost

While many self-driving companies continue to refine and optimise their technology through ongoing trials on public and private roads, the limitations of such grounds (both physical and regulatory) place agitating constraints on the ability of developers to improve the intelligence and safety that autonomous vehicles can achieve at higher speeds and dynamics.

Less conventional approaches to self-driving vehicle development are critical to enabling safe autonomy upwards of 100 kph, for applications such as cross-country logistics, medical care, disaster relief or security. The Indy Autonomous Challenge (featured in Issue 46) is one example of this in which self-racing open-wheel cars have been trained to manoeuvre and pass each other on track at over 200 kph.

The cost and logistics of fielding a team in that competition, however, may restrict its benefits and capacity for commercial spinouts to a handful of university and research departments worldwide. For others, a lower-cost option for safe testing of fast and dynamic self-driving tech, usable somewhere outside the restrictions of real-world roads, is sorely sought after.

To that end, the XTMV-800 ‘Nano MK.II’ UGV began in 2025 to serve a number of users hoping to put new autonomous tech and operating concepts to the test (XTMV standing for ‘Cross Tracked Modular Vehicle’). The small robot is the brainchild of Dr Christian Berth, lecturer at Technische Universität Berlin (TU-Berlin), who also serves as CTO in the company MBDH, with the Nano being developed and manufactured within that company under the brand and spinout XTRACKS-ROBOTICS.

Although engineering any uncrewed system involves balancing a plethora of conflicting trade-offs in subsystems and capabilities, constrained by SWaP-C as well as limits on reliability, lifespan, supply chain resilience and more, three points in particular have guided Dr Berth’s design and iterations of the XTMV-800.

One is a modularity and openness that enables incorporation of subsystems for a wide range of research projects, including end users focused on both road and indoor operations, as well as both civilian and defence applications.

Further advantages are driving stability and off-road capability. The Nano has a roller track chassis; a system that, according to Dr Berth, may encounter natural limitations (such as ground conditions and grain size) when scaled down to smaller vehicle sizes, but enables extremely stable and dampened movements that can be maintained for long periods of time, even in extreme terrain such as snow-covered hills.

Finally, XTRACKS-ROBOTICS has focused on keeping the Nano as low-cost and simple to manufacture as possible, such that units rarely cost more than a few thousand Euros to produce – an approach spurred not only by Dr Berth’s own background in education, but also as a tank officer for the German military.

“Most of the time, if there’s a serious accident or disaster in Germany like flooding or major fires, battalion commanders are made to give up their heavy support vehicles, such as those with cranes and other equipment, even though those vehicles are not specially constructed for such tasks, and doing so may force the whole battalion to stand still,” Dr Berth says.

“Having a civilian form of vehicle like this means you can avoid the need for heavy armour, and still drive stably while carrying heavy loads through damaged roads, difficult terrain or water of certain depths. Adding modularity and a robust electric powertrain would then also mean you can incorporate or replace all kinds of different payloads. So, instead of investing heavily into one specialised vehicle that you use once a year at most, you can use a single open vehicle for all kinds of situations.”

***The UGV is designed to be modular and open such that any subsystem or peripheral may be compatible***

As an open system, the Nano can integrate different companion computers, AI algorithms, sensors and other payloads as needed for virtually any application, enabling the capacity for military swarms of Nanos in defence applications (including mounting drone-in-a-box solutions onboard for combined, mass UGV-UAS operations), in place of otherwise more expensive and limited singular vehicles.

Henceforth, as well as its proven applicability as a tool for education and technical research, the Nano is well-positioned to serve in a variety of further group- and mesh-based operations, including perimeter security at sensitive and vulnerable locations such as airports, indoor security for industrial facilities, and monitoring or inspection in dangerous applications where lives of first responders may be saved through use of uncrewed tools to map out hazardous areas, such as wildfires or tsunami inundation areas.

From big to small

Around five years ago, Dr Berth began witnessing new designs for tracked vehicles coming from the private sector, and saw immense untapped potential for their use in other applications and with key engineered qualities such as those discussed above.

Preliminary construction ensued, not of a prototype Nano, but one for a significantly larger, optionally-crewed vehicle roughly 5 m in length (now designated the XTMV-5300) and capable of seating 2–6 people along with considerable cargo and subsystem payloads inside.

“There were several applications I had in mind for that, but one in particular was the devastating wildfires in California at the time, and others since, where firefighting personnel became encircled by fires and killed; with a tracked, optionally- crewed vehicle, teams could drive up to the limits of a safe location, disembark, and then pilot it forwards remotely or autonomously to gather mapping or thermal data,” Dr Berth muses. “Fallen trees with a circumference of up to 80 cm pose no problem for electrically powered, air-independent vehicles, even in smoke and fire conditions.”

After his colleagues noted the challenges with building a large vehicle from a blank sheet, however, the project pivoted towards a scaled-down version, now known as the XTMV-1600 (‘Mini’). From how highly functional it was relative to its construction cost (and their collective academic background), the group quickly recognised the Mini’s additional worth as a university research asset, particularly for exploring mobile- or cascade-type operations in which it might carry and launch drones, both then forming nodes in a network for first responders.

“It exhibited great climbing and stability, and the electric powertrain made it very flexible to drive and recharge indoors, but it was still a 120 kg prototype,” Berth recounts. “That’s too heavy for academics; impractical to carry around and deploy easily, and there was always the risk that you might run over an inattentive student.”

The Mini was thus reduced in size once again, resulting in the Nano prototype, which was made from several laser-cut and bent aluminium plates, screwed together and sealed.

Together with a number of other improvements, this led to an initial production series of 10 units for educational and research purposes, with two more units quickly following after the major research group TNO (Netherlands Organisation for Applied Scientific Research) requested them.

“The project specifications at TNO required relatively high endurance, and sufficient power had to be provided for the onboard computers, which resulted in a much greater overall mass than originally planned. I knew immediately that they would need more power,” says Dr Berth.

After switching to more powerful brushless DC motors and a reduction ratio of 1:3.5, a speed of around 40 kph and a gradability of more than 60° were achieved with a 12S Lipo. This performance tempted another buyer of the up-powered Nano MK.I with the new powerful drive to take it for a very fast ride in challenging terrain. This resulted in damage to the aluminium axles, prompting Dr Berth and his colleagues to switch to stainless steel for the axles and reinforce other chassis components and fuselage parts.

“We’d chosen aluminium, naturally, to make it light – but that clearly needed improvement in the opposite direction – so we put stainless steel axles and other parts in, and adjusted the damping system with stronger springs,” Dr Berth says.

“I’d already started considering a new version of the Nano, and so these changes were incorporated into the Nano MK.II, which also has a larger internal volume and suspension than the MK.I.”

Nano layout

Two battery packs integrate in the hull above the tracks (one on each side), the integration running in either 6S2P or 12S1P, depending on whether endurance or power, respectively, is to be prioritised. Electric motors and gears are integrated centrally in the back of the UGV, with extra space reserved on either side where additional batteries (among other components) can be placed if needed.

A plethora of other sections are dedicated component mounting spaces. The upper midsection of the Nano provides a payload bay with 20 L internal volume, accessible via an opening measuring 420 x 180 mm on top of the vehicle (and additional maintenance openings) to enable large, application-specific subsystems to be installed. At the front is a hatch, under which is a recess ideally placed for integrating stereo cameras for driving, gimbal systems as well as antennas.

“Antennas can also be placed atop the back of the vehicle, and in applications requiring more precision, we’ll aim to install differential satellite navigation [DGNSS] antennas and receivers atop the battery sections of the hulls, potentially up to four of them, inside the grey plastic covers there,” Dr Berth says.

On the Nano’s front face is a long, horizontal LED system that functions together with the back facing LEDs as a warning light, turn signal, brake light or battery and system state indicator, but future customisations can also use that recess to mount short-range Lidars or radars for project budgets that enable these.

“We have a tool shop for custom-producing small numbers of UGVs, but if someone wanted 100 or more units, then we have external suppliers for milling, drilling and other sheet metal processes who we’d go to instead, which would likely put the production price per unit around 1000 Euros,” Dr Berth adds. “Theoretically, at that rate, we could make 28,000 of these for the price of one Leopard 2A8 tank.”

Inside the Nano, a specialised Raspberry Pi or ESP32 is typically used as the main computer, given the capacity for easy integration and communication with other devices, and sufficient processing capability to handle command of the motor controllers and simple automated tasks, with ROS-2 serving as middleware. The company has also integrated Intel UPs and NVIDIA TX2s, AGXs and Orins as companion computers for work requiring higher computing power and AI functions.

***The high stability afforded by the drivetrain design suits the UGV towards highly dynamic manoeuvring and testing***

“For specific end-user applications, like defence, security or first response, we’d integrate the customers’ preferred payloads, comms and so on ourselves, and offer the open, plain version specifically for educational and research organisations,” Dr Berth explains.

Track limits

Aside from a few minor optimisations that remain, XTRACKS-ROBOTICS is largely satisfied with the Nano Mk.II’s physical and performance specifications, as well as its operating behaviours, with most of the improvements versus the first prototypes having gone into the drivetrain, especially the tracks.

“Most troublesome about the prototype Nano was that there was a particular incidence of the tracks either jumping or coming off of the wheels,” Dr Berth says.

“This also occurs with other tracked vehicles – there are similar cases with the Leopard 2, for example, when steering manoeuvres are performed on difficult terrain without sufficient power to the tracks – but we want end users to have a smooth, error-free experience with the Nano, as far as we can control it from our side.

“So, the new track has yet to come off the Nano at any point in its tests or demonstrations, even after our most recent trial where we pushed the system in -8 C temperatures, on very snowy, uneven grounds, including 45° climbs, for a little under 90 uninterrupted minutes. As long as no particularly large rocks or chunks of ice get inside the drive-sprockets, and users don’t put absurdly heavy payloads on that cause the tracks to become less tense, it drives like hell.”

Notably, that trial also investigated traction efficiency improvements, by integrating the Mk.II Nano UGV with the old track design on one side, and the new track on the opposite. Based on the difference in traction and power efficiency, the battery powering the drivetrain half with the old track was at a 14% SoC, while the new track had taken just 70% of its battery’s energy, leaving a remaining 30% SoC.

“As the Nano was originally built for the TU-Berlin, the development budget was constrained, and so I looked into some of the cheapest possible options for the tracks on the original prototype – after researching what kinds of tracks would fit the Nano at that time, we found tracks from a Chinese-made model of the Tiger 2 tank,” Dr Berth recounts.

“They were somewhat chain-like, made from plastic, and used small needles as the axles of each tread. I designed a sprocket that fitted into those tracks, as it was the most cost-efficient and fast way to keep the project moving.”

Additionally, a small, geared, brushed DC motor integrated at the drive wheel, enables 18 kph speeds on that vehicle with sufficient torque and speed for the application.

“But they were cheap motors, with only a small shaft; that, combined with a low-quality bearing inside the gear system we used at the time, meant we were putting extremely high tension on the sprocket over time, generating a bending force on it that pulled it away,” Dr Berth says.

Additional bending forces from the mismatch of traction subsystems caused fatigue and warping of the sheet metal hulls over time. Initial compensatory measures included adding some extra connection solutions into the fronts and backs of the drivetrains to keep the angles and shape of the tracks more consistent. Once testing indicated that these approaches resolved the issues, Dr Berth ordered sheet metal for the aforementioned initial production batch of the Mk.Is.

“But even though the problems with the bending forces weren’t affecting us anymore, in demonstrations that followed, the cheap-tracked treads would still come off, with some of the needles inside the tracks even coming out due to wear on occasion. So, at that point, we decided it was finally time to design a proper track system of our own,” Dr Berth continues.

Some inspiration was taken from pre-existing vehicle tracks, to design a track system consisting of a single, continuous piece of rubber, moulded using a proprietary two-component rubber solution that XTRACKS-ROBOTICS mixes and injects in-house.

“It took some experimentation to get a design with good manufacturability and performance, including one early attempt where we tried making the track as a long mould form, after which we’d glue the ends together to make it into a continuous, looped track – but it didn’t adhere or perform so well,” Dr Berth muses.

“So now, instead, we have a rounded mould, assembled out of three parts that fit together, into which we inject the rubber mixture and form the tracks. The material and equipment yield a very high-quality system with no porosity in the rubber, partially thanks to the mould parts being constructed with a good distribution of relief holes that enable the rubber to fill all the spaces it needs to for the right geometry and pressure.”

***A central payload bay enables the installation of payload sensors and companion computers, with data links and navigation-aiding systems ideally installed in compartments around the perimeter***

In addition to lacking the many gaps that the first tracks featured, the new generation tracks in the Mk.II Nano are also designed with larger teeth and guides that function to further prevent the chances of the tracks coming off of the UGV’s wheels.

Drivetrain evolution

For those unfamiliar with the concept, continuous tracks usually consist of parts such as drive sprockets (that transfer the driving force to the tracks and thus provide traction), support rollers (to minimise friction losses and abrasion), spring-loaded and damped road wheels (that distribute the weight of the vehicle over a large area with the help of the chains, thereby considerably minimising the specific pressure on the ground compared with systems using wheels alone), and an adjustable idler wheel (for deflecting and adjusting the track tension).

In the XTMV-800 Nano chassis, suspension is achieved by six wheels on each side – twelve in total – each of which is attached to a suspension arm equipped with double springs and dampers to control the movement of the road wheels. In addition, there are three support wheels that run at the top of each chain and a guide wheel at the front.

***A single BLDC electric motor drives each track via a 5-to-1 differential gear***

Customers of the German company occasionally refer to the UGV as a ‘tank’, which is undoubtedly because of its comparable tracked undercarriage. In contrast to simple tracked vehicles, which are often found in civilian applications, the classic support roller chassis used in modern battle tanks has many advantages and enables significantly higher speeds – even in demanding terrain – greater hull stability (and thus reduced susceptibility to tipping) and generally much better off-road capability.

“The inspiration from the road wheel suspension based on John Walter Christie’s designs offers the best way to achieve high agility, comfort for the vehicle hull and internal subsystems and, overall, the ability to drive both fast and stable,” says Dr Berth.

“Generally, a configuration like this might have provided more value, in terms of power and efficiency, if the Nano had been a larger vehicle – more akin to our larger products that will come in time – and maybe going with four wheels and one track per side would have made for a less complex and still very effective design. But we definitely would not have had such a stable platform as we do now.”

As mentioned earlier, each track is powered by a brushless DC motor. XTRACKS-ROBOTICS has moved away from brushed motors after recognising that commercial customers require greater payload capacity, even for smaller systems such as the Nano UGV. To deliver sufficient power, the motors for the UGV are potentially oversized, running up to 3 kW output power.

Dr Berth notes that this is a model used in certain heavy electric skateboards and similar personal e-mobility vehicles.

Each electric motor runs into a mechanically simple, 5:1 reduction, single-speed gearbox as standard. With that gear system, a 230 Kv BLDC motor drive and a 6S battery, the Nano moves at an operating speed of 16 kph. That nominal speed increases to 32 kph when combined with a 12S battery, and to 42 kph when the motor is upgraded to a 300 Kv model. XTRACKS also anticipates up to 65 kph if a further switch to a 3:1 gear ratio is made.

“After developing the new powertrain arrangement, I sent a kit of it to TNO so they could implement it as an upgrade replacing their brushed motors, and owing to the limits on space inside the MK.I Nano, the design included a belt drive connection between the parallel motors and gears,” Dr Berth adds.

“But the MK.II is larger. So, in this newer version of the Nano, there’s enough space for the motor and gear to integrate directly together on the inner walls of the UGV. That reduces parts, complexity and points of potential failure.”

Energy for the Mk.II Nano’s drivetrain comes from a pair of 5800 mAh LiPo battery modules, one powering each electric motor. As mentioned, this enables the UGV’s 90 minute typical endurance without additional onboard systems.

However, Dr Berth assumes that this duration will be revised upwards, possibly by up to two full hours, when a new wheel bearing for the road wheels is introduced in the MK.II.

“Previously, we used lubricated plain bearings because they were an inexpensive option – and if someone wanted to use a group of Nanos as attritable or disposable military vehicles, then we’d keep using them – but now we’re using ball bearings, which perform much better and last longer,” Dr Berth says.

“The wheel bearings were designed so that the existing suspension arms from the Nano MK.I and MK.II could continue to be used. Only the road wheels themselves had to be adapted to the new wheel bearings. However, this was achieved without much effort by using additive manufacturing processes. The sprockets and support rollers, which were also manufactured using additive manufacturing, could thus be produced quickly, easily and cost-effectively and then optimised during testing.

“Thanks to the first customers of our systems, such as TNO, Dronivo GmbH, TU-Berlin and many other users, as well as our own test series, we were able to gain a wealth of experience and invest it in the further development and improvement of our UGVs.”

From small to big

In addition to a few last optimisations to the MK.II and future customised versions as requested, XTRACK-ROBOTICS is now hard at work on the XTMV-1600 (or Mini MK.II), while also keeping track of how the larger XTMV-4000 and an XTMV-5200 UGV (slated for disaster response and civil security) designs might be refined and prototyped in future.

Much of the technology in the Nano MK.II can be scaled up to suit its larger siblings, such as drive controls and autonomy hardware and software. Much of the track design and geometry, similarly, is being ported into the Mini MK.II, albeit with some modifications.

***The company plans to take the lessons from the Nano MK.II and scale them up for developing its larger UGVs and optionally crewed ground vehicles***

“The Mini’s track can’t be formed as a single continuous piece because the required mould would be too big. So, we’re using multiple pieces made from rubber-coated aluminium, with the profile of the rubber still forming the teeth and guide lobes, and then we connect them together with the axles,” Dr Berth says.

“It took a while to prototype, but it’s ready and shows great improvements over the MK.I Mini. We’re also putting in a new PCB-based power distribution system, and we’re replacing much of the old cable harnessing for better stability and reduced weight.”

Improved electric motors and engineering simplicity have also been key targets for the Mini MK.II, with plans to start testing the new prototypes in real-world conditions imminently, with development of the company’s larger-still UGVs to follow thereafter.