By default the wilcox.test() function will calculate exact p-values if the samples contain less than 50 valid values and there are no ties in the values.
If the samples contain less than 50 finite values and there are ties in the values, the output produces a warning that an exact p-value cannot be obtained when the data contain ties (i.e. identical observations have the same rank). One method to deal with ties is to use so-called mid-ranks. This can be done by using the wilcox_test function from the coin package.
library(coin)
wilcox_test( MVPA_acc ~ Walkability, data=BEPAS, ties.method="mid-ranks")