Wheel slip
Odometry assumes rolling without slipping. Dust, carpet, collisions, and rapid acceleration break that assumption. Fuse another measurement when position accuracy matters.
Relate left and right wheel speeds to forward velocity, turning rate, and the path a differential-drive robot follows.
A simplified learning model. Values describe the model, not a connected robot.
Let vL and vR be wheel-ground linear speeds in metres per second, and b the distance between the wheels in metres. With forward-positive wheel speeds and left-positive rotation:
The experiment uses b = 0.18 m and plots a two-second ideal path. Equal wheel speeds produce a straight path. Opposite equal speeds rotate the robot in place. With both wheels reversing, the same equations still determine rotation; do not guess the turn from a forward-only rule.
With vL = 0.20 m/s and vR = 0.30 m/s, forward speed is 0.25 m/s and turn rate is about 0.56 rad/s. The ideal turn radius v/ω is 0.45 m. An encoder gives wheel angular speed; multiply by effective wheel radius to obtain linear speed.
Odometry assumes rolling without slipping. Dust, carpet, collisions, and rapid acceleration break that assumption. Fuse another measurement when position accuracy matters.
Measure effective wheel radius and track width under load. Small diameter differences create a persistent heading error even when encoder rates match.
Ramp speed commands to stay within motor torque and surface traction. Clamp both wheel commands consistently so a saturated motor does not distort the intended curve.
Given desired v and ω, calculate vL = v − ωb/2 and vR = v + ωb/2. Pass each wheel target to its own speed controller. A path planner chooses where to go; kinematics converts desired motion into wheel targets; feedback makes each wheel track those targets.
Manufacturer and project documentation. Reviewed 20 September 2026; check your exact board revision and software release.