Fail Forward: A chair undone is a chair well-done
This seminar focused on advanced subtractive manufacturing techniques, exploring the integration of 3-axis CNC milling and 6-axis robotic milling within a hybrid digital fabrication workflow. The process emphasized computational design-to-production strategies, including toolpath generation, material optimization, and robotic calibration. Through sequential stages of CNC roughing and robotic finishing, the seminar demonstrated how digital precision, tool accessibility, and material behavior can be coordinated to achieve complex geometries and high-quality surface refinement in contemporary fabrication processes.
Phase 1: Desing and CNC millling
We started by developing several design iterations and selected the one that best fit our concept. Then, we moved to the CNC phase, configuring the required parameters in the routing software, including toolpaths and material settings, to accurately translate the digital model into a physical prototype. Divided into Board 1 and Board 2. Board 1 uses a total of 1.73㎡ and takes 94.18 minutes, while Board 2 uses 1.11㎡ and takes 119.77 minutes.
Phase 2: Assembly Stage
Each layer of wood requires two sets of dowels, connecting the upper and lower templates, and glue is applied. During assembly, several issues arose:
- Tolerance for cutting depth: Each wood layer has a pre-cut hole for the dowel, but cutting 24mm depth does not guarantee the drill can pass through the board. Therefore, a tolerance must be added for cutting depth.
- Mismatch of dowels: For the 20-layer wood assembly, we could not find the correct upward-connecting dowels. Thus, we only used glue to join the layers, and later used screws to ensure tight fitting between layers before the glue dried. However, cracks still appeared in this layer during robotic milling.
Phase 3: Robotic Milling Stage / 6-Axis Robotic Finishing
- Patches: Patches are the division of our geometry into easier areas to work with, in this way we can facilitate the work of the machine. In our case we did some iterations, starting with 3 and up to 50 patches. With three patches we have many blue areas, with 15 it improves the appearance but in the hole you can see a blue zone, I 50 patches it improves a lot but it is an action that is not recommended, due to the amount of detail we are generating. To reduce the number of patches we change the height of the vector to 2.5, in which the almost complete reduction of all blue areas can be visible. The height of the vector helps us change the size of the cluster and in this way we can have better results with a reduced number of patches.
- Robotic milling strategy: We categorized all patches, including Bumps, Back, Top, Front, Seat, Seat Front, Leg Front, Feet Bottom, Feet interior, Bottom, Hole back, Hole front. We prioritized milling based on position and difficulty. Total machining time: 154.87 min.
- Kuka slot:
- Issues encountered: During simulation, we considered the geometry of the completed chair. In reality, the Kuka robot faced stacked wood, which is larger than the final chair geometry, causing potential collisions and material damage. Especially during milling of patch 19, the SP extra rotation on the Y-axis must be precisely between 20°- 40°, otherwise the robot could damage the chair back or seat.
- Possible solution: Divide patch 19 into four parts, and use different angles for the separated patches.
Lessons Learned and Summary
Check simulations – iterate early
Templates and file sharing protocols
Understand tool limitations
More glue!
