Why wheels win on flat ground
If you want a robot to move across a smooth floor, the simplest, cheapest, and most energy-efficient answer is almost always a wheel. This is the heart of wheeled locomotion: instead of lifting and placing feet, the robot rolls. A rolling wheel barely fights gravity at all — once it is turning, only a little friction and the motor's effort keep it going. That is why shopping carts, cars, and office chairs all roll rather than walk.
Compare that to a leg. A walking robot has to constantly catch itself from falling, swing a heavy limb forward, and balance on one foot at a time. All of that takes clever control and a lot of energy. Legs earn their keep on stairs, rubble, and rough wilderness, but on a warehouse floor they are overkill. A beginner building their first robot almost always starts with wheels, because you can get something moving in an afternoon.
Differential drive: two wheels and a trick
The most common layout for a small robot is the differential drive. Picture two wheels mounted on a left and a right side, each turned by its own motor, plus a free-rolling caster (a swivel wheel like the ones on a shopping cart) just to keep the robot from tipping. There is no steering wheel at all. You steer purely by choosing how fast each driven wheel spins.
The whole behavior of the robot falls out of one simple idea: compare the two wheel speeds. If both wheels spin forward at the same speed, the robot drives in a straight line. If the right wheel spins faster than the left, the robot curves toward the left, tracing a gentle arc. And here is the magic trick — if you spin the left wheel backward while the right wheel spins forward at the same rate, the robot turns on the spot, pivoting around its own center without going anywhere. That ability to rotate in place is something an ordinary car cannot do.
- Both wheels, equal speed, same direction → drive straight.
- Wheels at different speeds, same direction → curve along an arc (the slower wheel is on the inside of the turn).
- Wheels at equal speed, opposite directions → spin in place around the center.
Counting turns to guess where you are
Once a robot can move, the next question is: where is it now? The cheapest answer reuses the very wheels that move it. On each wheel's axle sits a rotary encoder, a small sensor that counts how far the wheel has turned — often hundreds or thousands of ticks per revolution. If you know the wheel's circumference, then counting ticks tells you how far that wheel has rolled.
Combine the left and right distances with the geometry of the previous section, and you can continuously update an estimate of the robot's position and heading. This running estimate built from wheel motion is called odometry — literally, measuring the road. It is a form of dead reckoning: figuring out where you are by adding up every small movement since a known starting point, the same way a sailor once tracked a ship using its speed and elapsed time.
left_dist = ticks_left / ticks_per_rev * wheel_circumference right_dist = ticks_right / ticks_per_rev * wheel_circumference forward = (left_dist + right_dist) / 2 # how far the center moved turn = (right_dist - left_dist) / wheel_base # how much the heading changed
Where simple two-wheel robots shine — and stop
Differential drive is everywhere precisely because it is cheap and capable. Robot vacuum cleaners use it to thread between chair legs and pivot in tight corners. Research and teaching labs lean on small two-wheel platforms because the motion is easy to model and program, making them the perfect sandbox for learning navigation. Many warehouse robots that scoot pallets around use the same scheme at a larger scale.
But the same simplicity sets the limits. The robot cannot move sideways, so squeezing into a parallel-parking-style gap means a clumsy back-and-forth shuffle. On a slick floor or in a sharp turn the wheels can slip, and slip is invisible to the encoders, quietly corrupting the odometry. And of course, the moment the floor turns into stairs or boulders, wheels give up — which is exactly where legs, the subject of a later guide, take over.