cor(x, y) measures how closely two columns follow a straight line, from -1 to 1. Near 0 means no straight-line pattern, not "no relationship". A missing value makes cor() return NA; use = "complete.obs" leaves those rows out.
A linear model describes one column with others. lm(minutes ~ distance_km, data = routes) reads "minutes explained by distance". coef(model) gives the intercept (the predicted minutes at zero km) and the slope (extra minutes per km). Pull one out by name: coef(model)[["distance_km"]].
Add predictors with +: in minutes ~ distance_km + stops, the stops coefficient is the extra time per stop for routes of the same distance. That is why it can differ from the slope of a model with stops alone.
predict(model, newdata = tibble(distance_km = 12, stops = 6)) gives the model's estimate for a new route. Correlation and coefficients describe the data; on their own they don't prove that one thing causes another.
A correlation, a model with two predictors, its coefficients, and a prediction.