Portugal, Santarém District: Czech researchers test TAROS autonomous military resupply missions during ARTEX 2026 near Constância
A Czech unmanned ground vehicle has demonstrated autonomous resupply capabilities during a multinational military exercise in Portugal, completing repeated missions while a human operator supervised its progress instead of continuously controlling its movement directly.
Researchers from the Faculty of Electrical Engineering at the Czech Technical University in Prague, alongside the University of Defence and VOP CZ, tested the six-wheeled TAROS platform during ARTEX 2026 in Portugal.
The exercise ran from September 28 to October 9 under a European Defence Agency initiative focused on unmanned ground systems. The trials were designed to assess autonomous technologies in demanding conditions resembling potential military operations.
During the exercise, TAROS travelled more than 100 kilometres, with several dozen kilometres completed autonomously. The platform repeatedly demonstrated that it could follow assigned routes, negotiate difficult terrain and transport supplies without continuous steering from an operator.
The results represent a step in the development of autonomous logistics vehicles, particularly for missions where reducing the need for personnel near exposed areas could be useful. Researchers stressed, however, that human supervision remains central to the system.
FEL ČVUT is developing the autonomous navigation system used by TAROS, a vehicle manufactured by Czech defence company VOP CZ. In the configuration tested in Portugal, the platform can carry a payload of up to 600 kilograms.
Rather than manually steering the robot throughout a mission, an operator specifies a destination and can add intermediate navigation points when required. TAROS then determines a suitable route using information collected from its surrounding environment under human supervision.
The navigation system relies on LiDAR, a laser-based sensing technology that measures the geometry of nearby terrain. It helps identify obstacles and select paths through challenging areas, including uneven ground, vegetation and natural obstacles safely.
Researchers repeatedly tested routes measuring about 2.5 to 3 kilometres. TAROS travelled on unpaved tracks and across terrain containing dense vegetation and natural obstacles, reaching autonomous speeds of up to 8 kilometres per hour.
The tests were intended to move the technology beyond controlled laboratory conditions and provide practical evidence about performance. Operating around vegetation, uneven surfaces and obstacles gives researchers a clearer picture of autonomous navigation when terrain becomes less predictable outdoors.
One important feature is the vehicle’s ability to recognize when it cannot safely continue. Instead of attempting to force a route, the system is designed to stop and allow the operator to intervene and take control.
That approach reflects the researchers’ emphasis on supervised autonomy rather than completely independent operation. The robot handles routine navigation, while a human remains available to make decisions when circumstances fall outside the system’s safe operational limits.
Czech Army Captain Štěpán Konecký said practical military use would depend on reliability, predictable behaviour and effective cooperation between the operator and the autonomous platform. Completing a route successfully, he indicated, is only one part of the wider challenge.
The exercise also provided an opportunity to examine autonomous driving after dark. Because LiDAR measures surrounding geometry using laser signals rather than relying on daylight, TAROS was able to navigate several hundred metres during nighttime field trials.
Nighttime testing is particularly relevant to autonomous ground vehicles because visibility can change dramatically after sunset. The ability to sense terrain independently of conventional lighting can support operations when human drivers would face reduced visual information available.
The supply missions were conducted in cooperation with the Portuguese army units. In the first scenario, TAROS was remotely controlled along a paved road shared with ordinary vehicles before switching to autonomous travel on a rougher parallel track.
That second route was primarily used by armoured vehicles, providing a more demanding environment than the paved section. The vehicle demonstrated that its navigation system could operate away from normal established roads while maintaining the assigned mission objective.
According to the Czech technical university, all subsequent supply missions were completed autonomously, with the operator primarily supervising the vehicle. The distinction is important because the operator was monitoring the system rather than continuously providing steering commands.
The trials also examined another potential military application: casualty evacuation. In a separate scenario, a Portuguese army volunteer travelled approximately 2.5 kilometres on a stretcher mounted to TAROS, allowing researchers to assess the platform’s suitability.
That evacuation journey was remotely controlled rather than autonomous. The test allowed researchers to examine how the platform could transport a person over terrain and assess practical factors such as the ride experience and suitability of the configuration.
Together, the supply and evacuation scenarios showed the broader range of tasks being considered for unmanned ground vehicles. Logistics and casualty transport can place personnel and conventional vehicles in situations where remote or autonomous systems may offer alternatives.
Jan Faigl, head of FEL ČVUT’s Computational Robotics Laboratory, said field exercises such as ARTEX provide information that cannot be obtained through laboratory testing alone. Feedback from potential military users can also influence future technical development.
The TAROS navigation project involves researchers and students from several departments at the Czech technical university. Their work combines robotics, control systems, sensing and computing to develop an autonomous platform capable of operating in complex outdoor environments.
ARTEX 2026 also allowed the Czech team to compare its approach with other unmanned ground systems. Researchers noted that autonomous navigation in terrain remained relatively uncommon among the platforms demonstrated, with many systems relying mainly on remote teleoperation.
The wider ARTEX programme brought together emerging defence technologies and military users at the Santa Margarida Military Camp in Constância. Its integration with the European Defence Agency campaign created a setting for multinational experimentation and operational feedback.
The Czech team will now use the exercise findings to improve autonomous driving, the operator interface and safe operation in difficult terrain. The work remains part of research and testing, rather than representing an announced operational deployment or procurement decision.
The Portuguese exercise therefore provides a practical milestone for TAROS, demonstrating supervised autonomy in resupply and additional testing in nighttime and evacuation scenarios. Further development will determine how reliably the technology can perform across changing military environments.