The automated flexible transfer line (AFTL) is designed with flexibility, reconfigurability and reliability to satisfy the requirements in real manufacturing environment. It contains multiple stages in series performing the assigned operations, and each stage consists of multiple machining cells with one robot and multiple identical machines. A multi-objective robust optimisation problem (MOROP) based on AFTL balancing problem under uncertainty with three conflicting objectives, i.e. minimise the expected line cycle time, minimise the probability of real line cycle time exceeding the expected line cycle time and minimise the smoothness index, is proposed in this paper. Three new scenario-based robust dominance (SRD) criteria are proposed, and two novel methods, i.e. heuristic based on branch and bound (HBB) and heuristic based on artificial bee colony (HABC), are designed. Different sizes of experiments on MOROP are made and solved by the methods, and the performances of HBB and HABC are tested against considered problems with different scenarios based on the SRD criteria. Overall results indicate that HBB is quicker in searching solutions and HABC is better in result quality, and both heuristics provide robust solutions for the AFTL balancing problem.