Then introduce accumulations, relating the velocity (just shown by an arrow, perhaps on even a map) to predictions about where the actor will end up. At this stage, 'velocity' is simply a label for the arrow, which shows 'how fast and in what direction', referring back to the groundwork you laid earlier on.
Then segue into a series of predictive stories, starting with a sequence of velocities and predicting the displacements.
For each step of the story, you can focus on the detail of the accumulation, varying the velocity and duration of the step, to vary the accumulated displacement. Iterating across the steps is the crux of 'reasoning with arrows': accumulation, which is just addition, is done by placing arrows tip to tail.
You could look at successive intervals, seeing how changing velocities changed displacement.
As these stories of motion can be described as track segments, as represented in the laboratory records, you can bring together the experimental records of motion with a predictive model: