Direct answer: Robotic die-casting tending should be planned as one continuous route rather than as a reachable pick point. The robot must approach on a usable gripping face, hold the hot part, leave the die area without interference, change pose only after it is clear, and release into a receiving position that the next operation can use. Pace is evaluated after that route works.
Who this is for: Process engineers, automation managers and equipment buyers planning robot tending around a die-casting machine, a hot part and a downstream transfer point.
Scope: This EVST guide uses footage of a robot approaching a die-casting machine, gripping a metal part, clearing the machine envelope and changing pose before placement. It treats those visible actions as path evidence, not as proof of cycle time, gripping force, heat resistance, quality yield or a completed customer deployment.
Robotic die-casting tending starts at the gripping face
The first useful drawing is not the robot reach circle. It is the part at release, the available gripping faces, the die opening and every fixed surface around them. A wrist can reach the nominal pick coordinates while the fingers, tool body or part edge cannot pass through the same corridor. The gripping face also decides the approach direction and the wrist orientation, so it needs to be settled before a robot model is selected. In practice, a few extra millimetres of tool thickness can remove the only clean approach that existed in a simplified model.
The same evidence is required whether the cell is described as machine tending or material handling. The robot machine tending overview provides adjacent context, but the application decision remains local: a representative part, a real end tool and the exact machine envelope have to be combined in one check. According to ISO 10218-2:2025, an industrial robot application is assessed as an integrated system, which is why the end tool, workpiece and surrounding machinery belong in the same review rather than in separate vendor drawings.
Part-family variation should be made visible at this point. A rib, handle, gate remnant or alternate cavity can change the only legal contact face and the swept volume without changing the nominal product name. The review therefore overlays the worst variants rather than validating the easiest sample and calling the family complete. If one variant needs a different grip or release orientation, that difference becomes a controlled recipe with its own proof, not an operator adjustment discovered after installation.
Approach and grip are only the first boundary
A successful pickup does not prove that the part can leave. The exit is usually the narrowest part of the route because the robot now carries the workpiece outside the end-tool envelope. A hot sprue, handle, flange or flash line may project toward a surface that was clear during the empty approach. The practical test begins at confirmed grip, then moves the complete swept volume along the first straight clearance segment. Turning early is avoided unless the model and a physical trial show that every part feature remains clear.
This is also where recovery logic becomes tangible. If the line stops with the part inside the die area, operators need a known state from which motion can be resumed or made safe. According to ISO 12100:2010, risk reduction follows a structured assessment of hazards and operating situations. The useful application record therefore includes normal exit, failed grip, incomplete release, power interruption and manual recovery, not only the smooth sequence shown during commissioning.
Change pose after the machine is clear
Once the complete part clears the die-casting machine, the robot can use more of its posture envelope. That is the correct place to rotate the casting for cooling, inspection, trimming or downstream placement. Combining exit and reorientation into one dramatic move may look efficient, but it consumes posture margin inside the most constrained space and makes collision reasoning harder. A separated route gives each segment a single purpose: clear the source, change pose, then travel toward the receiver.
The pose change has another consequence: gravity loads the grip in a new direction. A jaw that was stable while the part hung vertically may carry a different moment after the casting is tilted. This does not justify claiming a gripping force from footage. It identifies the load cases that must enter tool calculation and trial. Selection evidence comes from part mass properties, thermal condition, permitted contact points and the worst orientation, not from the robot payload label alone.
Thermal behaviour also affects the route indirectly. Heat can change acceptable hose routing, sensor choice, jaw material and how long the tool may remain near the open die. These are engineering inputs, not reasons to invent a temperature from colour or appearance. A design review records the customer’s declared thermal range, confirms component ratings and then observes representative operation. That chain lets the team distinguish a path problem from a component-life problem when a trial exposes a weakness.
The drop position belongs to the same path
Tending is incomplete until the tool releases and leaves the destination without disturbing the part. The receiving position has to accommodate the casting, the open fingers and the escape direction. It also needs to put the part in a pose that a conveyor, cooling rack, inspection station or trimming fixture can accept. A convenient empty corner for the robot may create a second handling step for the rest of the line.
The large workpiece pose and reach test is useful adjacent reading because it treats access as a changing set of robot postures rather than one radius. The same reasoning applies here on a smaller object: the least favourable wrist pose can occur at the drop, not at the die. An application review should show the entire swept volume, the tool-open state and the next system’s ownership of the part before it calls the route complete.
A decision table before any cycle-time promise
The planning sequence below keeps path evidence separate from performance assumptions. Each row has a document or trial that can close it, and each unresolved row remains an explicit hold instead of becoming an optimistic number in a sales estimate. That separation makes later acceptance easier because the team can trace each result to the input that created it.
| Decision | Evidence required | Hold condition |
|---|---|---|
| Gripping face and tool | Part drawing, contact restrictions, thermal state | No stable legal contact |
| Approach and first exit | Machine envelope and swept-volume check | Tool or part crosses a fixed boundary |
| Pose change | Worst-orientation load and robot posture | Grip or wrist margin is unverified |
| Drop and release | Receiver geometry and release trial | Part cannot be left and cleared |
| Cycle time | Timed representative run with handshakes | Path or interface states remain open |
Acceptance evidence that footage cannot supply
The reference footage shows a coherent approach, pickup, exit and pose change. It does not disclose the actual casting temperature, end-tool holding margin, machine signal sequence, measured cycle time or reject rate. None of those should be inferred from motion that appears smooth. According to ISO 13849-1:2023, safety-related control functions require defined performance and validation; a visible stop or guard does not establish that result by appearance.
EVST treats acceptance as a separate evidence package. A representative part set is run through normal and abnormal states, the machine handshake is observed, the grip is checked in the least favourable pose, and the downstream release is repeated. Project inputs and acceptance records then support a selection decision. Video remains useful for understanding the process order, but it is not substituted for measurements or the customer’s risk assessment.
The final review should keep a traceable route record: the model revision, end-tool revision, part variants, verified clearance cases and the exact trial condition. When a fixture or downstream receiver later moves, the affected segments can be rechecked without pretending the original acceptance covers a new geometry. This is especially valuable in tending cells, where a small machine-side change can alter both normal motion and recovery while leaving the robot program name unchanged.
Frequently asked questions
Should robot reach be checked before the gripper is designed?
No. The gripper and part extend the swept volume and determine the wrist attitude, so they must be included in the reach and collision check.
Why not rotate the casting while it leaves the die?
Rotation may be possible, but it consumes space in the tightest part of the route. A clear exit followed by reorientation is easier to validate and recover.
Can the video be used to estimate cycle time?
No. It omits machine handshakes, thermal constraints, confirmation delays, abnormal states and the full downstream exchange.
What is the first representative trial?
Use the least favourable part orientation and a production-intent end tool, then prove approach, clear exit, pose change and release as one route.
Project inputs for an application review
A useful application review begins with the objects and interfaces that define the route, not with a preferred robot model.
- Part envelope, mass properties, temperature and release pose
- Permitted gripping faces and surface restrictions
- Machine opening, fixed obstacles and signal list
- Destination geometry and required drop orientation
- Normal, recovery and acceptance scenarios
Send those inputs to EVST for a path-first review of the end tool, robot posture, machine clearance and receiving position. Related reading: headstock-tailstock positioner guidance.