⚡ electrical engineering

Multi-Axis Additive Manufacturing for Customized Automotive Components

This paper proposes a variable exposure method for multi-axis Digital Light Processing (DLP) 3D printing that modulates UV illumination duration based on local layer thickness to eliminate the need for subdividing non-uniform layers, thereby significantly reducing print time and support structure waste in the fabrication of complex automotive components.

Original authors: Uzair Aziz Muhammad, Zheng Liu

Published 2026-04-15

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Uzair Aziz Muhammad, Zheng Liu

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are trying to build a complex, organic-shaped sculpture out of a special liquid plastic that hardens when hit by light. This is essentially what 3D printing does, specifically a high-tech version called DLP (Digital Light Processing).

Usually, 3D printers work like a very precise, slow-moving cookie cutter. They shine a light on a flat layer of liquid, harden it, lift the platform up a tiny bit, and repeat. This works great for boxes and cylinders. But try to print a curved, bumpy car part (like a custom dashboard or a sleek bumper), and you run into a problem: gravity and overhangs.

The Problem: The "Cookie Cutter" Limitation

In traditional printing, if you have a part that sticks out sideways (like a T-shape), the printer has no way to hold it up while it hardens. To fix this, the software adds scaffolding (support structures).

The Analogy: Imagine building a sandcastle with a wide overhang. You have to build a temporary wooden frame underneath it to hold the sand while it sets. Once the castle is done, you have to break the frame away, which is messy, wastes wood, and leaves rough spots on the sandcastle.
In 3D Printing: These "scaffolds" use extra resin, take time to print, take time to remove, and often ruin the smooth finish of the final part.

The Solution: The "Dancing Robot"

This paper introduces a smarter way to print: Multi-Axis Printing. Instead of keeping the print bed flat, they attach it to a robotic arm.

The Analogy: Imagine a potter throwing a pot on a wheel. Instead of the clay staying flat and the potter's hands moving up and down, the potter tilts the wheel to match the curve of the clay. The robotic arm tilts the print bed so that the "light" always hits the surface at a perfect 90-degree angle, no matter how curvy the part is. This means you often don't need any scaffolding at all!

The New Challenge: The "Uneven Cake"

Here is where the paper's specific innovation comes in. When you tilt the bed, the layers you are printing are no longer flat pancakes; they are curved slices.

The Problem: If you shine a light on a curved slice, one side of the slice might be very thin, while the other side is very thick.
The Old Way: If you shine the light for the same amount of time everywhere, the thin part gets overcooked (burned or deformed), and the thick part gets undercooked (it doesn't harden enough and falls apart).
The Traditional Fix: You could chop that one curved layer into 10 tiny, flat layers to make the thickness uniform. But that makes the print take 10 times longer!

The Paper's Innovation: The "Smart Dimmer Switch"

The authors developed a method to fix this without slowing down the print. They created a "Cure Map."

The Analogy: Think of the printer's light source not as a single on/off switch, but as a smart dimmer switch with thousands of tiny individual knobs (one for every pixel of the image).
How it works:
1. The computer looks at the curved layer.
2. It calculates: "This spot is thin, so give it a quick, dim flash of light. That spot is thick, so give it a long, bright flash."
3. It generates a grayscale image (like a black-and-white photo) where bright white means "shine bright and long" and dark gray means "shine dim and short."
4. The printer flashes this image onto the resin. The thick parts get enough energy to harden, and the thin parts get just enough to harden without burning.

Why This Matters for Cars

The title mentions "Automotive Components." Car parts are full of curves, curves, and more curves.

Speed: Because they don't have to split layers into tiny pieces, the printing is much faster.
Cleanliness: Because the robot arm tilts to avoid overhangs, they need almost no "scaffolding." This saves money on expensive resin and means the final car part comes out smooth and perfect, ready to use immediately.
Customization: This makes it easy to print custom car parts for specific drivers or unique designs that would be impossible to make with traditional molds or standard 3D printers.

In a Nutshell

This paper teaches a 3D printer how to tilt its head to follow the curve of a car part, and then teaches its light bulb how to adjust its brightness pixel-by-pixel so that every part of the curve hardens perfectly at the same time. It's like upgrading from a flashlight that only has an "On" and "Off" button to a sophisticated stage lighting rig that can paint the perfect amount of light on every inch of a complex sculpture.

1. Problem Statement

The paper addresses a critical limitation in Digital Light Processing (DLP) 3D printing when applied to multi-axis systems for automotive component fabrication.

The Context: Multi-axis DLP printing uses robotic arms to reorient the build platform, allowing for the fabrication of complex, organic geometries without the wasteful support structures required by traditional fixed-axis (planar) printing.
The Challenge: When the build platform is dynamically reoriented, the resulting sliced layers are non-planar. Consequently, the thickness of a single layer varies spatially across its cross-section.
The Limitation: Conventional DLP systems apply a uniform exposure time and intensity to an entire layer. In multi-axis printing, this uniform approach fails:
- Over-curing: Thinner regions receive too much energy, leading to dimensional inaccuracy.
- Under-curing: Thicker regions receive insufficient energy, causing inter-layer adhesion failures and structural weakness.
Existing Solutions: Previous attempts to solve this involved subdividing non-uniform layers into multiple thinner, uniform sub-layers. However, this drastically increases the total layer count, print time, and the need for support structures, negating the efficiency benefits of multi-axis printing.

2. Methodology

The authors propose a variable exposure method that generates spatially resolved cure maps, allowing a single non-uniform layer to be printed accurately without increasing the layer count. The methodology consists of a two-stage pipeline implemented in Python:

A. Curved Iso-Surface Slicing Algorithm

Adapted from a multi-axis FDM framework (Fang et al.), this stage decomposes a 3D triangle mesh ( $M$ ) into a sequence of non-planar contour layers.

Scalar Field Construction: A scalar field $\phi(x, y, z)$ $ϕ (x, y, z)$ is constructed over the mesh vertices. Two modes are supported:
- Warped- $z$ field: Introduces sinusoidal perturbations to the vertical coordinate to create curved layers.
- Radial field: Generates concentric spherical layers around a user-defined center point.
Iso-Surface Extraction: The scalar range is divided into $N$ intervals. Each layer $k$ is extracted as an iso-surface at a specific value ( $iso_k$ ).
Contour Generation: Edge crossings are computed via linear interpolation and chained into closed polyline contours ( $P_k$ ). The algorithm enforces minimum/maximum thickness bounds to ensure operability.

B. Per-Pixel Cure Map Generation

Once a layer contour is obtained, a grayscale exposure mask is generated for the DLP projector.

Physical Model: The process relies on the Beer-Lambert law (Jacobs working curve), which relates cure depth ( $C_d$ ) to photoexposure dose ( $H_e$ ):
$C_d = D_p \ln\left(\frac{H_e}{H_0}\right)$
Where $D_p$ is the resin penetration depth and $H_0$ is the critical exposure dose.
Inverse Calculation: The system inverts this relationship to solve for the required grayscale command $G(u, v)$ $G (u, v)$ at each projector pixel $(u, v)$ $(u, v)$ based on the local target thickness $h(u, v)$ $h (u, v)$ :
$G(u, v) = \left\lfloor \frac{255 \cdot H_e}{I_0 \cdot t_e} \exp\left(\frac{h(u, v)}{D_p}\right) \right\rfloor$
- $I_0$ : Peak irradiance.
- $t_e$ : Exposure time.
Mask Encoding: The layer contour is rasterized into a binary map. For interior pixels, the required grayscale value is computed and clamped to [0, 255]. Pixels outside the contour are set to 0.
Quality Metric: A saturation ratio ( $\sigma_k$ ) is calculated (ratio of pixels at 0 or 255) to detect if the target thickness exceeds the material's curing capabilities, signaling a need for recalibration.

3. Key Contributions

Variable Exposure Framework: Introduction of a method to modulate UV illumination duration/intensity proportionally to local layer thickness, eliminating the need to subdivide non-uniform layers.
Spatially Resolved Cure Maps: Development of an algorithm that converts 3D non-planar contours into 2D grayscale masks that encode precise UV doses for every pixel, ensuring accurate curing depth across varying geometries.
Efficiency Optimization: The approach reduces the total number of layers required for complex parts, directly translating to faster print times and reduced material waste.
Low-Overhead Extension: The method is a software-based extension to existing multi-axis DLP pipelines, requiring no additional hardware modifications.

4. Results

The pipeline was validated using representative 3D models, including a handle with delicate surface structures and a T-shaped object (geometrically impossible to print without supports in fixed-axis DLP).

Curved Slicing: Successfully generated non-planar contour stacks (e.g., 200 layers for the handle) that follow the object's geometry.
Cure Mask Visualization: Generated grayscale masks (PGM images) where pixel intensity correctly correlated with local thickness. Brighter pixels indicated regions requiring deeper cure (thicker sections), while darker pixels indicated thinner sections.
Support Structure Reduction: A comparative analysis against Preform (a standard planar slicing tool) demonstrated significant improvements:
- Handle Model: 47% reduction in support volume; layer count reduced from 3,176 (Preform) to 100 (Proposed).
- T-Shape Model: 97% reduction in support volume; layer count reduced from 290 (Preform) to 100 (Proposed).
- Material Savings: Significant reduction in resin consumption due to the elimination of support structures.

5. Significance

This research provides a practical solution to a major bottleneck in multi-axis DLP printing, making it viable for the automotive industry.

Customization: It enables the rapid prototyping and reproduction of complex, customized automotive components with organic shapes that were previously difficult or impossible to fabricate without supports.
Manufacturing Efficiency: By reducing print times (fewer layers) and material waste (less support), it lowers the cost and lead time for high-value automotive parts.
Geometric Fidelity: The method ensures high dimensional accuracy and surface quality by preventing over/under-curing, which is critical for functional automotive components.
Future Impact: The work lays the groundwork for fully automated, end-to-end fabrication pipelines integrating robotic path planning with advanced photopolymerization control.