In this talk we consider a complex scheduling problem which arises in liquid handling robot scheduling in a slide staining instrument for cancer diagnostics. The problem can be considered as generalised scheduling problem due to the presence of almost all constraints considered in the scheduling literature and some additional new constraints such as “on-demand jobs” and “delayed” sequence dependent set-up time. The problem is a NP-hard, large scale, on-line scheduling problem with hard constraints. Finding a robust, efficient algorithm is becoming one of the main priorities of the manufactures of the instrument. We present a heuristics algorithm, which improves productivity of the instrument, reduces costs and increases reproducibility of the tests. Moreover, the scheduler can be used in designing of other instruments with the best resource utilisation outcomes. The algorithm is tested on real world data sets.