Unmonitored move instruction

Is it somehow possible to issue a move instruction in a way that any subsequent move instructions would override the previous move instruction?

For example I assign the robot to move to coordinates (100,0). But during this move a more important move instruction arrives (-100,0). In this case I would like to abandon the current move instruction and start the new move as fast as possible. 

The new move might be in a completely opposite direction of the current move and might be issued while not even 10% of the previous move was completed.

This is for a little real time computer vision object tracking application. Our algorithm tracks an object and estimates where the robot should go to follow it. However if the tracked object is suddenly moved to the opposite direction the robot first tries to finish the previous move instruction instead of instantly starting the new one.

To summarize I would like to execute a move instruction as fast as possible independent of the state of previous move instruction that is currently executing. And I would like to repeat this unlimited number of times.

So is there something similar?

Kind regards,

Laov