A set-membership approach to consensus problems with bounded measurement errors.