Blog

What Is an Unmanned Ground Vehicle? Definition, Types and Real Applications

Tom Cosgrove,

Reading Room: The Fictionlab Blog

TL;DR: An unmanned ground vehicle (UGV) is a mobile robot that operates on the ground without a human on board. It carries sensors, computing, and payloads, and it is controlled remotely, semi-autonomously, or fully autonomously. UGVs can be described by drive system, mass class, and autonomy level, and they are used in research, inspection, agriculture, and education.

What Is an Unmanned Ground Vehicle and How Does It Differ From Other Robots?

An unmanned ground vehicle is a wheeled, tracked, or legged machine that moves across terrain while the operator stays elsewhere – at a desk, in a control room, or nowhere at all if the system runs autonomously. The defining characteristic is not autonomy. It is the absence of an onboard human. A teleoperated pipe inspection crawler and a self-navigating field robot are both UGVs.

That distinction matters because the term gets used loosely. Three elements are usually present in any practical unmanned ground vehicle:

  • A mobility base – drivetrain, suspension, and chassis that determine where the robot can physically go.
  • A perception and compute stack – cameras, LiDAR, IMU, encoders, and an onboard computer running software such as ROS 2.
  • A payload interface – mechanical mounting points, power rails, and data buses that let you attach the sensor or manipulator that actually does the job.

These elements determine how useful and adaptable a UGV is, but not every UGV includes all of them to the same extent. A teleoperated UGV may have limited onboard compute, while a basic platform may have little or no dedicated payload interface. The mobility base alone is a chassis, not a complete robot.

Why Terminology Confusion Costs You Time

Vendors, research papers, and standards bodies use overlapping vocabulary. “Mobile robot” is the umbrella term. “UGV” emphasizes ground operation and unmanned control. “AMR” (autonomous mobile robot) usually refers to a robot with onboard navigation, often for indoor logistics and fleet-managed operations. “Rover” carries planetary-exploration connotations and often describes rugged outdoor UGVs with high ground clearance.

When you are specifying a platform, ignore the label and read the specification sheet. Ground clearance, ingress protection rating, payload capacity, and software stack tell you far more than the acronym.

How Are Unmanned Ground Vehicles Classified by Drive System?

Locomotion determines where an unmanned ground vehicle can operate and how much energy it burns getting there. The main categories differ sharply in terrain capability, mechanical complexity, and maintenance burden.

  • Differential-drive wheeled platforms – two or more independently driven wheels, steering by speed difference between sides. Mechanically simple, low part count, and easy to model in software. This is a common architecture in research-grade mobile robots, including the Leo Rover, which uses four driven wheels with differential (skid) steering.
  • Ackermann-steered platforms – car-like steering geometry. Better rolling efficiency at speed, but a non-zero turning radius means they cannot rotate in place.
  • Tracked platforms – continuous tracks distribute weight over a larger contact area, which helps on loose sand, mud, and snow. The trade-off is higher drivetrain friction, more wear components, and harder odometry.
  • Rocker-bogie and articulated suspensions – passive linkages that keep all wheels in ground contact over obstacles. Common on planetary-analog rovers where climbing irregular rock fields matters more than speed.
  • Legged platforms – quadrupeds and hexapods that step over discontinuities instead of rolling across them. They can handle terrain that defeats wheeled systems, but have high control complexity and generally lower payload efficiency than comparable wheeled platforms.

For most research and inspection work, a differential-drive or Ackermann wheeled unmanned ground vehicle covers the requirement. Tracks and legs become justified when the terrain genuinely defeats wheels, and that threshold is higher than many teams assume.

How Suspension and Clearance Affect Field Performance

Two numbers provide useful early indicators of outdoor behavior: ground clearance and wheel diameter. Clearance affects the risk of the chassis high-centering on obstacles. Wheel diameter influences the step height the robot can climb, together with traction, wheel placement, approach angle, and suspension design. A platform with large wheels relative to its body, such as the Leo Rover’s 130 mm diameter wheels on a compact chassis, handles grass, gravel, and curb-like steps better than a low-slung platform of similar mass.

Check the official specification for both figures before committing to a deployment site. Visit the site with a tape measure if you can.

How Do You Classify UGVs by Mass and Payload Class?

Mass class drives nearly every practical decision: transport logistics, safety procedure, battery sizing, and whether a single person can lift the robot into a vehicle. There is no single universal standard, but the field generally works with these informal bands.

  • Micro and desktop class (under approximately 5 kg) – educational platforms and swarm research units. Usually indoor-only, limited payload, short runtime.
  • Portable class (approximately 5-25 kg) – often manageable by one person, subject to local lifting guidance. This band covers most research rovers and lightweight inspection platforms. The Leo Rover sits here, with a documented mass of approximately 6.5 kg and a nominal payload capacity of 5 kg.
  • Mid class (approximately 25-150 kg) – often requires two people or a ramp. Supports manipulator arms, larger sensor suites, and multi-hour missions. The Raph Rover falls into this band and is built for heavier payloads and more demanding outdoor terrain.
  • Heavy class (above approximately 150 kg) – vehicle-scale UGVs for agriculture, construction, and site logistics. Typically requires suitable transport equipment, documented safety procedures, and often a formal risk assessment before operation.

Payload capacity is the number that actually constrains your project. A sensor head with LiDAR, a stereo camera, a GNSS receiver, and a small manipulator can approach or exceed 4 kg, particularly after mounting hardware and cable management are included. If the platform is rated for 5 kg, you may be near the limit before those additions are accounted for.

What Are the Levels of Autonomy in an Unmanned Ground Vehicle?

Autonomy is a spectrum, and most deployed unmanned ground vehicle systems sit somewhere in the middle rather than at either extreme. A workable breakdown follows the control relationship between operator and machine.

  1. Direct teleoperation – the operator issues continuous velocity commands over a radio or network link and watches a live video feed. The robot generally makes no navigation decisions beyond low-level motor control and any essential stabilization.
  2. Assisted teleoperation – the robot overlays safety behaviors on operator input, such as obstacle-triggered speed limiting or collision avoidance. The human still steers.
  3. Waypoint autonomy – the operator defines goal poses; the robot plans and executes paths using SLAM or GNSS localization. The human supervises and intervenes.
  4. Task-level autonomy – the operator defines a mission (“inspect these 12 points, return to dock”); the robot sequences navigation, data capture, and recovery behaviors itself.
  5. Persistent autonomy – the system runs continuously, including self-docking, fault recovery, and schedule management, with human attention only on exceptions.

Moving one level up multiplies software effort. This is why ROS 2 matters so much in practice. The Nav2 stack, the ros2_control framework, and the tf2 transform system give you tested implementations of path planning, controller management, and coordinate bookkeeping rather than writing them yourself.

Why ROS 2 Became a Common Software Layer

ROS 2 replaced the central ROS Master architecture of ROS 1 with DDS-based communication and discovery, added configurable quality-of-service policies per topic, and introduced lifecycle-managed nodes. For an unmanned ground vehicle operating over lossy Wi-Fi, QoS settings let you mark camera streams as best-effort while keeping command and safety topics reliable.

Both Fictionlab platforms ship with ROS 2 support and open documentation, which means ROS 2-compatible software developed for one robot can often transfer to another with configuration changes rather than a complete rewrite. That portability is a practical argument for the ecosystem, and one reason many university labs choose it.

How Does a UGV Compare to an AMR and a Drone?

Teams often evaluate an unmanned ground vehicle against an indoor AMR or an aerial platform. The comparison below reflects typical characteristics rather than absolute limits, since specific models vary widely.

  • Operating environment: UGV – often outdoor or mixed terrain, including unstructured areas. AMR – commonly structured indoor floors, warehouses, and hospitals. Drone – open airspace, limited by obstacles and regulation.
  • Endurance: UGV – typically hours on a single charge. AMR – typically hours with automated docking. Drone – typically tens of minutes for multirotors.
  • Payload-to-mass ratio: UGV – high, since lift is not required. AMR – high. Drone – low; every gram costs flight time.
  • Sensor stability: UGV – good, with vibration from terrain. AMR – very good on flat floors. Drone – often requires active stabilization for high-quality imaging.
  • Regulatory overhead: UGV – generally site-level safety rules, though public-space and road use can involve additional requirements. AMR – industrial safety standards for shared spaces. Drone – national aviation authority rules, pilot competency requirements, or certification in many jurisdictions.
  • Close-range inspection: UGV – strong, can park and dwell for long periods within battery and mission limits. AMR – strong indoors. Drone – excellent for elevated structures, limited dwell time.

The practical rule: if the target is on or near the ground and the task needs dwell time, an unmanned ground vehicle is usually the lower-friction choice. If the target is elevated, inaccessible from the ground, or spread across a broad area, aerial coverage can be more effective. Many field programs run both and fuse the data.

What Are the Real Civilian Applications of Unmanned Ground Vehicles?

Common civilian UGV applications include four areas where moving a sensor across ground repeatedly is the core requirement.

Research and Algorithm Development

University and institutional labs use an unmanned ground vehicle as a reproducible testbed for SLAM, multi-robot coordination, reinforcement learning, and planetary-analog navigation. The requirement here is openness: documented interfaces, accessible source code where applicable, and the ability to swap sensors without vendor approval. Fictionlab publishes hardware and software documentation for this reason, and the research applications page covers typical lab configurations.

Infrastructure and Industrial Inspection

Routine inspection of substations, tunnels, culverts, solar farms, and process plants involves walking the same route with the same instruments. A UGV carrying a thermal camera, gas sensor, or acoustic sensor converts that into a repeatable automated pass with consistent viewpoints. Consistency is the real value – comparing thermal images from closely repeated poses week over week can make trends easier to detect than in less standardized handheld surveys.

Agriculture and Environmental Monitoring

Under-canopy operation is where ground platforms can beat aerial ones. A small unmanned ground vehicle drives between rows, measures soil moisture, counts plants, and images leaf undersides – viewpoints that drones often cannot safely or effectively reach. Environmental monitoring applications include forest plot surveys, glacier and volcanic terrain sampling, and shoreline mapping, using wheeled, tracked, or legged platforms selected for the terrain and environmental conditions.

Education and Training

Robotics curricula need a platform students can break, reconfigure, and debug. A compact UGV with ROS 2 support lets a course move from teleoperation to SLAM to autonomous navigation within a semester, on the same hardware, with simulation in Gazebo supporting development and testing alongside the physical robot.

What Should You Check Before Selecting a UGV Platform?

These checks surface most of the problems that appear after a platform arrives.

  1. Payload budget with margin. Sum sensor mass, mounting hardware, cables, and any manipulator. Compare against the official rating, not a marketing figure.
  2. Power availability at the payload connector. Check voltage rails and continuous current, not just peak. LiDAR and onboard GPUs draw more than expected.
  3. Ingress protection for the actual site. An IP rating covers specified levels of dust and water protection, not corrosive atmospheres or temperature extremes.
  4. Software stack and version support. Confirm which ROS 2 distribution the vendor supports and how long it stays supported.
  5. Mechanical interface documentation. You need dimensioned drawings or CAD files to design brackets. Ask before buying.
  6. Spare parts and repair path. Field platforms break. Find out whether wheels, motors, and controllers are orderable separately.

Frequently Asked Questions

What is an unmanned ground vehicle in simple terms?

An unmanned ground vehicle is a ground-based robot that operates without anyone riding on it. A human may control it remotely, or it may navigate on its own using onboard sensors and software. The category covers everything from small research rovers to vehicle-scale agricultural machines.

Is a UGV the same as an autonomous robot?

No. “Unmanned” means no onboard operator. “Autonomous” means the robot makes navigation decisions itself. A teleoperated inspection crawler is unmanned but not autonomous. Many unmanned ground vehicle systems operate across several autonomy levels depending on the mission phase.

What is the difference between a UGV and an AMR?

An AMR is typically an autonomous mobile robot built for structured environments such as warehouses, often with fleet management and floor-level safety sensors. A UGV is a broader category that includes outdoor platforms, teleoperated systems, and rugged terrain robots. In common ground-robot usage, AMRs are a subset of UGVs; not every UGV is an AMR.

Why do many research UGVs run ROS 2?

ROS 2 provides tested implementations of navigation, transform management, hardware abstraction, and inter-process communication, so research teams can spend more time on their actual contribution instead of infrastructure. It also makes drivers and algorithms more portable between platforms and supports simulation-to-hardware workflows.

How much payload can a small UGV carry?

It depends on the platform class. Portable research rovers in the 5-25 kg band typically carry single-digit kilogram payloads; the Leo Rover, for example, is rated for approximately 5 kg. Mid-class platforms such as the Raph Rover support substantially heavier sensor suites and manipulators. Always confirm against the manufacturer’s official specification.

Can a UGV operate in rain and dust?

Many outdoor platforms carry an ingress protection rating that covers specified levels of dust and splashing or spraying water. Check the exact IP figure in the official documentation, and remember that any payload you add has its own rating. A sealed robot carrying an unsealed camera is not a sealed system.

Do UGVs need special regulatory approval?

Requirements vary by jurisdiction and site. Ground robots generally face lighter regulatory overhead than drones, which are governed by aviation authorities in most countries. Operating on private property or a campus typically falls under site safety procedures; public-road or public-space operation can involve additional local rules. Verify with the relevant authority before deployment.

Where to Go Next

If you are scoping a research, inspection, or education project and need an open, documented, ROS 2-native unmanned ground vehicle to build on, review the full technical specification and payload interface details on the Leo Rover platform page.


Read more