It is often stated that a robot is more precise than accurate. Okay sure, but why?
While grossly over simplified, a robot is simply a rigid body with a bit of math. This would be sufficient, except for the fact we live in an analog world. To name a few factors, things like encoder resolution, calibration, deviations in the manufacturing/assembly process, and thermal cycles can impact accuracy and precision. You will have to forgive the exaggeration, but for the same reason you would never buy a true “to scale” solar system, some is required to prove the point.
Encoder Resolution: Regardless of an encoder’s resolution there are always discrete steps associated with an encoder. Assuming everything else is ideal, the larger the encoder step the less precise you can be. Decreased encoder resolution reduces precision while maintaining accuracy.
Thermal cycling: If we go back to an ideal robot. The perfect situation only happens at a particular temperature. Temperature fluctuations change the link dimensions and thus vary the location of the EOT. This is often negligible, nevertheless, it is something that contributes to a deviation from nominal.
Calibration: Back to the ideal robot one last time, now if the base joint is a couple degrees off all downstream links will be skewed relative to this inaccuracy. If we had infinite encoder resolution and a constant temperature the robot could be perfectly precise however it would but inaccurate. Said another way, assuming the same joint configuration, the robot can always get to a taught location but will be consistently in the wrong location.
Manufacturing/Assembly process: In the same way that a joint can be misaligned, all manufactured parts have some deviation. Determining the correct location based on the joints of a robot is based on propagating down the kinematic chain. Because the exact dimensions of a part vary between each robot it is not possible to know exactly where the robot EOT is just that it is within tolerance.
It is often stated that a robot is more precise than accurate. Okay sure, but why?
While grossly over simplified, a robot is simply a rigid body with a bit of math. This would be sufficient, except for the fact we live in an analog world. To name a few factors, things like encoder resolution, calibration, deviations in the manufacturing/assembly process, and thermal cycles can impact accuracy and precision. You will have to forgive the exaggeration, but for the same reason you would never buy a true “to scale” solar system, some is required to prove the point.
Encoder Resolution:
Regardless of an encoder’s resolution there are always discrete steps associated with an encoder. Assuming everything else is ideal, the larger the encoder step the less precise you can be. Decreased encoder resolution reduces precision while maintaining accuracy.
Thermal cycling:
If we go back to an ideal robot. The perfect situation only happens at a particular temperature. Temperature fluctuations change the link dimensions and thus vary the location of the EOT. This is often negligible, nevertheless, it is something that contributes to a deviation from nominal.
Calibration:
Back to the ideal robot one last time, now if the base joint is a couple degrees off all downstream links will be skewed relative to this inaccuracy. If we had infinite encoder resolution and a constant temperature the robot could be perfectly precise however it would but inaccurate. Said another way, assuming the same joint configuration, the robot can always get to a taught location but will be consistently in the wrong location.
Manufacturing/Assembly process:
In the same way that a joint can be misaligned, all manufactured parts have some deviation. Determining the correct location based on the joints of a robot is based on propagating down the kinematic chain. Because the exact dimensions of a part vary between each robot it is not possible to know exactly where the robot EOT is just that it is within tolerance.