Robotics Explained: Mobile Robots, Robot Arms, Hardware, and Software

Robotics brings together machines, sensors, electronics, mechanics, and software so a system can sense its surroundings and carry out useful actions. Some robots move through a building or across a field. Others stay in one place and handle objects with a mechanical arm. Many modern systems combine both: a mobile base carries an arm to the work.

This guide introduces the major families of robots, with a closer look at mobile robots and robot arms, then explains the hardware and software tools used to build them.

What makes a machine a robot?

A robot usually combines several capabilities:

Mechanical structure: a chassis, links, wheels, legs, joints, or other moving parts.
Actuation: motors or other mechanisms that create movement.
Sensing: devices that measure the robot’s state or environment.
Control and computing: electronics and software that turn goals and sensor readings into actions.
Power and communication: batteries or another power source, plus wired or wireless connections.

A robot does not have to look human, use artificial intelligence, or work without supervision. A factory arm following a programmed path is a robot. So is a small wheeled rover controlled remotely.

A practical map of robot types

Robots can be classified by where they work, how they move, and what they do. These categories overlap: a warehouse robot can be mobile, autonomous, and fitted with a manipulator.

Family	How it works	Common uses
Fixed industrial robots	A base is bolted to a work cell; the mechanism moves through a repeatable workspace.	Welding, painting, machine tending, palletizing, assembly
Mobile ground robots	Wheels or tracks move a platform over floors or terrain.	Warehouses, delivery, inspection, agriculture, research
Legged robots	Articulated legs step over uneven ground and obstacles.	Research, inspection, search and rescue
Aerial robots	Usually multirotor or fixed-wing aircraft with onboard control.	Mapping, inspection, photography, surveying
Underwater robots	Remotely operated or autonomous vehicles move underwater.	Marine science, inspection, exploration
Humanoid and social robots	Human-shaped or expressive systems interact with people or use human-oriented spaces.	Research, assistance, education, entertainment
Medical and assistive robots	Specialized systems help clinicians or support human movement.	Surgery, rehabilitation, prosthetics, mobility assistance
Collaborative robots (cobots)	Robots designed for flexible work near people, with safeguards chosen for the application.	Assembly, inspection, packaging, small-batch production
Mobile manipulators	A mobile platform carries a robot arm and tool.	Picking, inspection, replenishment, laboratory automation

These labels describe broad families, not guarantees of capability or safety. A robot must be selected and validated for its actual task and environment.

Mobile robots: movement, localization, and navigation

A mobile robot changes its position in the world. Its core challenge is to move toward a goal while estimating where it is, detecting hazards, and responding to changes.

Common mobile robot designs

Differential-drive: two powered wheels, often with one or more passive casters. Turning is achieved by driving the wheels at different speeds. It is a popular indoor design because it is mechanically simple.
Ackermann steering: wheels steer like a car. It suits higher-speed outdoor platforms, though it needs room to turn.
Omnidirectional or mecanum drive: specialized wheels allow sideways motion as well as forward and rotational motion. This is useful in tight indoor spaces, but the mechanics and control are more involved.
Tracked platforms: tracks provide traction on rough ground, at the cost of more friction and energy use on smooth floors.
Legged platforms: legs step over terrain and obstacles, but balance, control, and mechanical design are substantially more complex.
Aerial and underwater vehicles: these are mobile robots too, but their motion and sensing must account for flight or fluid dynamics.

AGVs and AMRs

An automated guided vehicle (AGV) generally follows a defined route or infrastructure, such as floor markers, reflectors, wires, or a fixed path. An autonomous mobile robot (AMR) commonly uses onboard sensing and software to localize itself and plan routes that can adapt to obstacles. In real products the boundary is not always sharp: check the specific robot’s navigation method rather than relying on the label alone.

How a mobile robot navigates

A typical navigation pipeline has these stages:

Read sensors. Wheel encoders estimate wheel rotation; an IMU measures angular motion and acceleration; cameras or LiDAR observe the environment. Ultrasonic or depth sensors can help measure nearby obstacles.
Estimate motion and position. Odometry estimates how the robot moved. Localization compares sensor observations with a map or other reference. All estimates have uncertainty and can drift.
Build or use a map. Mapping software represents free space and obstacles. Simultaneous localization and mapping (SLAM) estimates a map while tracking the robot’s position when a suitable prior map is not available.
Plan a route. A planner selects a path to the destination, while local motion control adapts to nearby obstacles and constraints.
Control the drive system. Controllers translate a desired motion into wheel or motor commands and use feedback to reduce errors.
Handle exceptions safely. A robot may stop, request help, re-localize, or choose another route if a passage is blocked or sensor data is unreliable.

Navigation quality depends on the whole system: sensor placement and calibration, wheel traction, lighting or surface conditions, map quality, compute capacity, and software configuration.

Where mobile robots are used

Mobile robots move materials between warehouse stations, inspect large facilities, transport laboratory samples, monitor crops, survey terrain, and support research. A mobile manipulator can bring an arm to shelves, workbenches, or machinery, extending what either a base or arm can do alone.

Robot arms: joints, reach, and manipulation

A robot arm, or manipulator, is a chain of links connected by joints. The tool attached at the end is often called an end effector. It might be a gripper, suction cup, welding torch, camera, screwdriver, or a specialized tool.

Each independently controlled joint contributes a degree of freedom (DOF). More degrees of freedom can give the arm more ways to position and orient its tool, but also increase the complexity of planning and control. A common industrial articulated arm has six rotary axes; smaller or specialized arms may have fewer or more.

Main arm geometries

Articulated arm: a series of rotary joints gives a broad, human-arm-like workspace. Used for welding, material handling, tending, and many general industrial tasks.
SCARA: a fast assembly-oriented design with a particular combination of horizontal compliance and vertical rigidity. Often used for pick-and-place and light assembly.
Cartesian or gantry: linear axes move along straight coordinate directions. Its box-shaped workspace is easy to understand and can be a good fit for large-area handling or precise repetitive motion.
Delta or parallel robot: several arms connect a fixed base to a moving platform. This architecture can move lightweight loads quickly, such as in sorting or packaging.
Collaborative arm: a category of industrial arms designed with features intended to support flexible human-robot work. The completed application still needs an appropriate risk assessment, tooling, speed limits, and safeguards.

How an arm reaches a target

A user may specify where the tool should go in Cartesian space (position and orientation). The robot controller or motion-planning software solves inverse kinematics to find joint positions that can reach that target. It then plans a path while considering joint limits, collisions, speed, acceleration, and the tool’s task.

The arm needs feedback. Joint encoders report motion; force/torque sensing can help detect contact or regulate interaction; vision can locate parts; and the controller coordinates the actuators. A gripper also matters: a delicate object, a heavy box, and a flexible cable require different gripping strategies.

Where robot arms are used

Robot arms perform repetitive or precise work such as welding, painting, assembly, packaging, machine tending, laboratory handling, inspection, and material transfer. The right arm depends on payload, reach, accuracy, speed, workspace, tooling, environment, and how people share the work area.

Hardware: the physical building blocks

Structure and motion

Frame, chassis, and links provide the mechanical structure.
Joints, bearings, wheels, tracks, and gearboxes guide and transmit motion.
Actuators create motion. Electric servo and brushless motors are common; steppers are useful in some lower-cost or open-loop applications; pneumatic or hydraulic actuators are used where their force, speed, or environment makes sense.
Motor drives deliver controlled electrical power to motors. A small DC motor may use an H-bridge driver; industrial servos use dedicated drives and feedback connections.
End effectors interact with objects: parallel grippers, vacuum tools, compliant fingers, welding tools, and many others.

Sensors

Encoders measure wheel or joint rotation.
IMUs measure acceleration and angular velocity, helping estimate motion and orientation.
Cameras and depth cameras support visual detection, measurement, and manipulation.
LiDAR measures distance to surfaces and is commonly used for mapping and obstacle detection.
Ultrasonic and infrared distance sensors can be useful for short-range obstacle sensing.
Force/torque and tactile sensors measure contact and help a robot interact with objects.
Electrical and environmental sensors monitor battery voltage, current, temperature, and other conditions.

No single sensor is perfect. Combining several sensor types can improve estimates, but it adds calibration, integration, and failure-handling work.

Compute, control, and power

A microcontroller can handle motor control and fast, predictable input/output tasks. A single-board computer or industrial PC can run higher-level perception, planning, user interfaces, and networking. Many robots use both, with clear boundaries between real-time motor control and general-purpose software.

Mobile systems need batteries, power conversion, wiring protection, and often a battery-management system. Industrial systems may use fixed power and dedicated controllers. In every design, power budget, heat, emergency stopping, and electrical protection are part of the engineering—not afterthoughts.

Software: from firmware to robot applications

Robot software is layered. Low-level firmware reads sensors and drives outputs. Control software regulates motors. Higher layers handle perception, maps, motion planning, task logic, user interfaces, logging, and communication with other systems.

Useful software tools

ROS 2 (Robot Operating System 2): open-source middleware and tools that help robot components communicate through messages, services, and actions. ROS 2 is not an operating system in the same sense as Linux or Windows; it is a robotics software framework that runs on a host operating system.
Nav2: a ROS 2 navigation framework used to build mobile-robot navigation behaviors, including planning and recovery.
MoveIt 2: a ROS 2 platform for manipulation, including robot-arm motion planning, kinematics, and perception integration.
Gazebo: a robotics simulator for testing robot models and environments. Simulation helps development, but it cannot prove real-world safety or replace hardware testing.
RViz: a visualization tool commonly used to inspect robot state, maps, sensor data, and planned motion in ROS workflows.
CAD and robot-description tools: CAD packages help design the mechanism; URDF and related formats describe robot links, joints, and sensors for ROS-based tools.
Embedded development tools: C/C++ toolchains, Python, microcontroller SDKs, debuggers, and serial tools support firmware and electronics integration.
Version control and testing: Git, code review, automated tests, simulation scenarios, and recorded sensor data make robot software easier to maintain and reproduce.

The right tool depends on the robot. A simple line-following rover may need only a microcontroller and a small control loop. A mapping robot may need a Linux computer, ROS 2, localization, and navigation packages. An industrial arm may use a vendor controller, a PLC, or an external planning framework integrated with the cell.

A beginner-friendly build path

Start with a small two-wheel mobile base, a microcontroller, a motor driver, and wheel encoders.
Learn safe power wiring, motor control, encoder feedback, and basic closed-loop speed control.
Add an IMU and a simple distance sensor; log what the sensors report before trying autonomy.
Build a small arm kit or simulated arm and learn how joint angles relate to tool position.
Explore ROS 2 and simulation once you understand the basic electronics and coordinate frames.
Add autonomy or manipulation in small steps, testing each change first in a safe workspace.

Keep the first projects low-energy and supervised. Pinch points, moving wheels, batteries, and unexpected motion can cause injuries or damage.

How to choose between a mobile robot and an arm

Choose a mobile robot when the task requires transporting the robot or payload between locations. Choose a fixed arm when work happens repeatedly within a defined cell and a stable base is suitable. Consider a mobile manipulator when both movement between work areas and object handling are necessary.

Before choosing, write down the task, payload, reach or travel distance, accuracy, speed, floor or workspace conditions, sensor requirements, human interaction, maintenance needs, budget, and safety constraints. This turns a broad technology choice into a concrete engineering problem.

Key idea

Robotics is not one machine type or one software package. It is the combination of mechanics, actuation, sensing, computation, and control for a particular job. Mobile robots focus on moving through the environment; arms focus on reaching and manipulating objects. Learning the hardware and software layers makes it easier to understand how they can work together.