Instead of motion accuracy, usually positioning accuracy is measured.
Positioning of a robot arm is relative to its fixed base. We can measure the relative angular positions of its joints very accurately using encoders and using a geometrical model of the robot, these measurements give an accurate estimate of the position of the end effector (or tool center point) of the robot. Encoders are relatively inexpensive and used in almost every industrial robot arm
In case of mobile robots there is no fixed base, there is no relatively inexpensive way of measuring its position relative to a fixed coordinate system in space. In some cases this fixed coordinate system is not even visible form the robot (e.g. in a cluttered room). There are very expensive systems which could measure the position if there is direct visibility (e.g. laser tracker systems) but this leads to other complications (line of sight visibility) and significant cost increase. If we could measure the position accurately, we could position the robots accurately. Instead robots use onboard laser scanner (inaccurate compared to laser tracker systems), cameras and-or other sensor systems and calculates its position relative to features in its environment, which are themselves sometimes inaccurately positioned on a map.
It all comes down to measurements. If we can measure accurately, we can usually also position accurately.