Let Xi be a random variable from a uniform distribution on the interval [0,b]. Thus the lower bound is known but not the upper bound.
You observe n many independent and identically Xi's, what is the expected value of the sample mean X_bar = (1/n) ⁿΣᵢ₌₁ Xi?
How could you transform X_bar to make it an unbiased estimator of b (that is to say the expectation of the estimator is equal to b)?