← Robot list · The mobile robot in 3D · Learn the basics
A mobile robot is a robot that is not fixed to the ground: it moves in its environment. The most important choice when you design one is how it moves. The wheels you pick decide which movements are possible, how well the robot carries loads, how well it knows where it is, and how hard it is to control. This page shows every common wheel type, then lets you drive four different drive types to feel the difference.
Engineers sort wheels by two questions: can the wheel steer, and can it slide sideways? Four classes are classic (fixed, steered, castor, omnidirectional), plus balls and tracks.
Pick a wheel below. It is a 3D model you can turn with the mouse. Then push it with the joystick (or let the demo run) and watch how it performs: how it rolls, where it slides, where it skids.
| Material | Good for | Limits |
|---|---|---|
| Polyurethane (PU) | indoor floors, quiet, long life, most AGVs and AMRs | hard floors only, can slip on dust |
| Rubber | grip, small and educational robots | wears faster, can leave marks |
| Pneumatic (air-filled) | outdoor, bumps, soft ground | can puncture, pressure changes the wheel radius |
| Solid or foam-filled | outdoor without punctures | heavier, less comfortable |
| Hard plastic or nylon | castors, light loads, low friction | little grip |
| Steel | robots running on rails | needs rails |
Siegwart and Nourbakhsh describe a robot with two numbers: the degree of mobility δm (how many motions the wheels allow right now, without steering) and the degree of steerability δs (how many wheels you can steer). The degree of manoeuvrability is δM = δm + δs. A robot with δM = 3 can reach any position and heading, but only a robot with δm = 3 can do it instantly: it is holonomic.
| Configuration | Wheels | (δm, δs) | δM | Holonomic? | Typical use |
|---|---|---|---|---|---|
| Differential drive | 2 driven fixed wheels + 1 or 2 castors | (2, 0) | 2 | no | indoor AMRs, vacuum robots, education |
| Tricycle | 1 steered and driven front wheel + 2 fixed rear wheels | (1, 1) | 2 | no | tow tractors, counterbalanced AGV forklifts |
| Car-like (Ackermann) | 4 wheels: 2 steered at the front, 2 driven at the rear | (1, 1) | 2 | no | self-driving cars, outdoor delivery robots |
| Synchro drive | 3 or more wheels, all steered and driven together | (1, 1) | 2 | no | some research robots |
| 3 omni wheels (kiwi) | 3 omni wheels at 120° | (3, 0) | 3 | yes | RoboCup, small holonomic platforms |
| 4 Mecanum wheels | 4 Mecanum wheels in an "X" layout | (3, 0) | 3 | yes | omnidirectional forklifts, warehouse platforms |
| Swerve drive | 3 or 4 wheels, each steered and driven | (2, 1) | 3 | after the wheels have turned | competition robots, heavy omnidirectional AGVs |
| Skid-steer / tracked | 4+ fixed wheels or 2 tracks, turning by slipping | (2, 0) | 2 | no (it slips) | rough terrain, rescue, agriculture |
Choose a drive type, press Start, and change the inputs. The big drawing shows the wheels, how fast each wheel turns (arrows and numbers), the steering angles, and the ICC (the point the robot is turning around). The arena below it shows where the robot goes.
A robot is holonomic if it can move in any direction at any moment, without first having to turn. A 4-Mecanum robot can slide sideways at once. It is the easiest to program for paths.
The robot has more positions to reach than motions available right now. A car cannot slide sideways, so to park beside a wall it must go forward, reverse and turn. Differential and car-like robots are nonholonomic.
Instantaneous Centre of Curvature: the single point of the plane that does not move at this instant. Every other point turns around it. For wheels that roll without slipping, all the wheel axles must meet at the ICC.
Track width is the distance between the left and right wheels. Wheelbase is the distance between the front and rear axles. A bigger track makes the robot more stable but turns more slowly.
| Environment | Examples | How they move |
|---|---|---|
| Ground | warehouse AMR, vacuum robot, rover, self-driving car | wheels, tracks or legs |
| Air | drones (UAV): inspection, delivery, mapping | propellers or wings |
| Water surface | survey boats (USV) | propellers, water jets |
| Underwater | ROV (remote-controlled), AUV (autonomous) | thrusters |
| Locomotion | Strong points | Weak points |
|---|---|---|
| Wheels | simple, fast, very energy-efficient on flat ground, cheap | poor on stairs, rubble and soft ground |
| Tracks | rough and soft terrain, stairs, low ground pressure | slip when turning, heavier, more energy, noisy |
| Legs | any terrain a human can walk on, steps and stairs | complex, slow, more energy, hard to control |
| AGV (Automated Guided Vehicle) | AMR (Autonomous Mobile Robot) | |
|---|---|---|
| Guidance | fixed path: magnetic tape, wire in the floor, reflectors, QR codes | free navigation with a map, lidar and cameras |
| Obstacle | stops and waits | plans a path around it |
| Change a route | modify the floor infrastructure | change the map or the software |
| Best for | stable, repeated routes with heavy loads | changing layouts, mixed traffic with people |
Proprioceptive sensors measure the robot itself. Exteroceptive sensors measure the world.
| Sensor | What it gives | Used for |
|---|---|---|
| Wheel encoder proprio | rotation of each wheel | odometry (distance driven, speed) |
| IMU: gyro + accelerometer proprio | turning rate, acceleration, tilt | heading, detecting slip and slopes |
| Motor current proprio | the load on a motor | detecting a blocked wheel or a collision |
| 2D lidar extero | distances to obstacles in a plane | mapping, localization, obstacle detection, safety scanner |
| 3D lidar extero | a 3D point cloud | outdoor robots, cars, 3D mapping |
| Camera, RGB or depth extero | images, depth, markers | recognizing objects and QR codes, 3D obstacles |
| Ultrasonic, infrared, time-of-flight extero | short-range distance | cheap obstacle detection, parking |
| Bumper, cliff sensor extero | contact, floor drop | last-resort safety, stairs |
| GNSS (GPS) extero | position outdoors (centimetres with RTK) | outdoor and farm robots |
| Line, magnetic tape, UWB, RFID extero | position on a guide or from beacons | AGVs, indoor positioning |
With encoders the robot adds up its small motions: x ← x + v·cosθ·dt, y ← y + v·sinθ·dt,
θ ← θ + ω·dt. This is dead reckoning. Errors never go away: wheel slip, a wheel
radius that is slightly wrong, uneven floors, and encoder resolution. A small heading error grows into a
large position error after a few metres. So robots correct odometry with other sensors.
map → localization → global planner → local planner → speed command → wheels
(A*, Dijkstra) (DWA, TEB, MPC) (v, ω)
| Field | Tasks | Typical drive |
|---|---|---|
| Warehouses and logistics | carrying boxes, shelves, pallets; picking | differential, Mecanum (forklifts) |
| Factories | moving parts between machines (AGV, AMR) | differential, tricycle, omnidirectional |
| Hospitals and hotels | delivering meals, medicine, linen | differential |
| Cleaning | vacuum and floor-scrubbing robots | differential |
| Outdoor delivery and cars | sidewalk delivery, self-driving vehicles | car-like (Ackermann), 4-6 wheels |
| Agriculture | crop monitoring, weeding, harvesting | tracks, 4-wheel steering |
| Inspection and rescue | pipes, tunnels, disaster zones, stairs | tracks, legs |
| Space | exploring other planets | rocker-bogie wheels |
| Education and research | learning robotics, competitions | differential, omni, Mecanum |
| Your need | Choose | Why |
|---|---|---|
| simple, cheap, indoor | differential + castor | 2 motors, can turn on the spot |
| tight spaces, sideways motion | Mecanum or omni | holonomic |
| high speed, outdoors, long distances | car-like (Ackermann) | stable at speed |
| rough, soft ground or stairs | tracks or rocker-bogie | grip and obstacle crossing |
| very heavy loads | tricycle or differential with big wheels | strong fixed wheels |
AGV automated guided vehicle · AMR autonomous mobile robot · DOF degree of freedom · IMU inertial measurement unit · lidar laser distance scanner · odometry position estimated from the wheels · SLAM build a map while locating yourself · PID a simple feedback controller · ROS Robot Operating System · BMS battery management system · kinematics the study of motion without forces · dynamics the study of motion with forces.