Plan the sequence.
See the overlap.

Parallel scheduling uses a deterministic priority list: repeatedly choose the earliest listed action whose dependencies have been ordered, then place it at the earliest available time after its dependencies finish while reserving all its resources for the full duration. Actions cannot pause or split. Resource capacity is one. Sequential scheduling uses the same dependency order and runs one action at a time.

This is a feasible priority-based schedule, not a guarantee of the shortest possible cycle. Reordering ready actions can change resource allocation. Resource occupancy is relative to this single modeled cycle. The dependency-only longest path and busiest-resource workload are lower bounds, not independent proofs of the active bottleneck. They exclude setup, transport and waits unless you explicitly add them as actions.

The example loads for 3 s, clamps for 1 s, drills for 6 s, prepares the next part for 4 s, inspects for 2 s and unloads for 2 s. Preparation can overlap drilling because its operator resource is separate. Parallel timing is 14 s versus 18 s sequentially. Add the fixture resource to preparation to see resource contention remove that saving.

Durations must be positive. Dependencies must exist and form an acyclic graph. Cycles, self-dependencies, missing IDs and invalid durations block results. All actions start at or after time zero; end/start contact is allowed without overlap. Times use unrounded values.

A timing diagram does not establish collision clearance, machine safety, interlock adequacy or PLC correctness. This conceptual single-cycle planner does not model machine dynamics, stochastic variation, buffers, pipelining or repeated-cycle production tracking. The optional motion editor supplies ideal rest-to-rest action times, not actuator sizing or a proven machine timing result.