In the rapidly evolving field of artificial intelligence, the demand for bionic robots in service industries has grown significantly. Beyond basic functional requirements, there is an increasing expectation for these bionic robots to exhibit rich emotional expressions during human-robot interactions. This naturally imposes stricter demands on the performance of robotic facial expressions. To address this, our team leveraged 3D printing technology to design and fabricate facial mechanical transmission systems for a bionic robot. We adopted an iterative “design-print-iterate-reprint” approach, continuously refining the model through trial and error to achieve an optimal product. This article summarizes the key technical aspects of the 3D printing process, highlights challenges encountered and their solutions, and demonstrates the entire workflow to provide reference for future mechanical designs. Our goal is to showcase the convenience of 3D printing and promote its broader adoption and development in robotics.
The bionic robot head we designed focuses on mechanical transmissions for components such as the eyes, eyelids, eyebrows, and mouth, which are critical for facial expressions. The product is constrained to use purely mechanical structures for motion, with over 80% of parts manufactured via 3D printing. The head frame must be securely fixed to provide stable support. Through iterative comparisons and adjustments of transmission mechanisms, part dimensions, and anthropomorphic aesthetics, we developed three versions, culminating in a third-generation optimal model. This final design enables eyelid opening and closing, eyebrow movement with small-angle rotation, eye lateral movement and rotation about a point, and mouth opening and closing.
The eyebrows are simplified as rigid rods, with their motion representing eyebrow deformation. For vertical movement, a crank-slider mechanism is employed, while a swinging block mechanism handles rotation, allowing for up-down motion and clockwise/counterclockwise rotation. The eyeball and eyelid are designed as an integrated assembly; the eyelid is a spherical shell structure housing the eyeball, controlled by a dedicated mechanism for rotation and eyelid actuation. The eyelid uses a parallel push-pull mechanism, and the eyeball incorporates a combination of parallelogram and swinging block mechanisms for coordinated movement. The mouth is divided into an upper jaw fixed to the skull and a lower jaw connected via a jaw joint, enabling rotation around it. A cam mechanism drives the lower jaw, with a pin connection allowing rotation; notches on the cam facilitate precise control for open, half-open, and closed positions.
The head structure is based on bionic principles, utilizing a three-dimensional design where the head is fixed to a frame for support. Each motion component is controlled by an independent mechanical structure located at the rear of the face, with defined motion ranges and secured via screws and nuts. The degrees of freedom for each part are summarized in the table below.
| Motion Unit | Mechanism Used | Number of Units | Motion Function | Motion Range | Total Degrees of Freedom |
|---|---|---|---|---|---|
| Eyebrow | Crank-Slider, Swinging Block | 2 | Vertical Motion, Rotation | 0–10 mm, 0–30° | 4 |
| Eyelid | Linkage Mechanism | 2 | Opening and Closing | Open/Close | 2 |
| Eyeball | Parallelogram, Swinging Block | 2 | Lateral Rotation, Central Rotation | 90° Rotation, 360° Spin | 4 |
| Lower Jaw | Cam Mechanism | 1 | Vertical Motion | Open, Half-Open, Closed | 1 |
For the 3D printing process, we selected a Tiertime UP 300 printer with UP Studio 3 software and ABS filament. Tools such as scissors, pliers, scrapers, brushes, tweezers, adhesive sticks, and grease were prepared for post-processing tasks like part removal, support cleanup, and deburring. To ensure strong adhesion and prevent warping, we applied a specialized adhesive to the build plate. Pre-printing steps included platform inspection, initialization, leveling, height adjustment, filament loading, and temperature setting.
Parameter settings were customized for each component. For the eye control assembly—the most complex part with numerous small transmission rods and supports—we optimized orientation to minimize supports, especially for the thin, curved eyelid. Placing the eyelids horizontally reduced external supports. To facilitate easy support removal, we used Vase mode for infill, Lines pattern for infill paths, and reduced bond strength to 20% in support settings. For the eyeball, after comparing four orientations, we chose one with minimal supports and good surface finish. Infill paths for the eyeball used Zigzag mode for top and bottom layers. The eye bracket, with its curved arm, was printed at Normal speed with Zigzag infill to ensure strength.
The head structure, serving as a protective shell, required durability and aesthetics despite its large, thin design. We set layer thickness to 0.2 mm for quality, enabled dense supports with three layers for curved sections, and used Random mode for start point optimization to enhance appearance. Internal infill employed Zigzag mode for strength, with adjusted density and support bond strength to minimize surface marks. Orientation was chosen to maximize stability and reduce supports, prioritizing surface smoothness.
For the eyebrow control mechanism, composed of gears, rods, and pins, we disabled supports and arranged parts flat on the build plate for efficiency. Infill paths used Zigzag mode to enhance stability, with Fast speed mode and Random start point optimization to avoid surface lines. The frame, acting as the base, demanded high strength to withstand forces. We applied cylindrical sub-model settings with increased infill density and extrusion ratio for the rods to improve bending resistance, while other parameters remained standard.
During printing, several issues arose. Initially, vertically oriented eyelid models were prone to damage during assembly due to weak layer adhesion. By reorienting them horizontally, we improved strength along the X-axis, eliminating breakage. Warping occurred due to high print speed and temperature, causing material stringing and edge lifting; adjusting these parameters resolved it. To optimize fit between pins and holes, we tested various hole sizes and found a 0.1 mm diameter difference provided the best fit. Adhesion failures were addressed by cleaning the platform and fine-tuning material temperature. Nozzle clogging, caused by excessive print speed leading to material backflow and solidification, was mitigated by reducing speed and cleaning the nozzle with tweezers after extrusion.
The final assembled bionic robot product demonstrated high integration, ease of assembly and disassembly, and met expected performance. It successfully replicated various facial expressions, enhancing the emotional interactivity of the bionic robot. For instance, the combination of mechanisms allowed for nuanced expressions such as surprise or happiness, showcasing the potential for improved human-robot interaction.
In conclusion, our bionic robot facial product was refined through iterative design and printing cycles. During modeling, we considered structural and functional requirements,预留合理的打印容差 for high-precision connections and seals. To reduce support material usage and post-processing time, we employed “bridging” techniques where possible. Parameter optimization was crucial for print success; we tailored settings for each component, experimented with various approaches, and incorporated manual calculations for support angles and material strength, ultimately achieving a method that balanced easy support removal, proper adhesion, and strength distribution. Printer maintenance, including platform leveling, nozzle height calibration, and regular lubrication, ensured optimal performance. Challenges like collapse, warping, stringing, and clogging were overcome through continuous adjustment. Post-printing, parts were inspected and processed by removing supports, bases, and burrs. The final prints met our targets for dimensional accuracy and surface quality.
This project enhanced our technical skills and provided a comprehensive understanding of 3D printing. The technology offers immense potential and benefits, warranting ongoing exploration and practice. For example, the use of 3D printing in bionic robot development allows for rapid prototyping and customization, which is essential for advancing human-robot interaction. The mechanical design principles applied here can be extended to other areas of robotics, such as limb movements or full-body expressions. Future work could involve integrating sensors for real-time feedback or using multi-material printing to simulate soft tissues, further blurring the line between machines and living beings. As 3D printing evolves, its role in creating more lifelike and interactive bionic robots will only grow, driving innovation in service, healthcare, and entertainment industries.
To delve deeper into the mechanical aspects, we can model some of the motions using basic equations. For instance, the crank-slider mechanism for eyebrow movement can be described by the relationship between the crank angle and slider displacement. Let \( \theta \) be the crank angle, \( r \) the crank radius, and \( l \) the connecting rod length. The displacement \( x \) of the slider is given by:
$$ x = r \cos \theta + \sqrt{l^2 – r^2 \sin^2 \theta} $$
This equation ensures precise control over the vertical motion of the eyebrows in the bionic robot. Similarly, for the cam mechanism in the jaw, the profile can be designed using parametric equations to achieve specific motion curves, enhancing the realism of expressions. The integration of such mathematical models with 3D printing parameters, like layer height and infill density, allows for optimized performance. For example, the infill density \( \rho \) can be related to the part strength \( S \) through empirical formulas such as \( S = k \cdot \rho^m \), where \( k \) and \( m \) are material constants. By iterating on these parameters, we achieved a balance between weight and durability in the bionic robot components.
In summary, the synergy between mechanical design and 3D printing has enabled us to create a highly functional and expressive bionic robot face. This approach not only accelerates development but also opens new possibilities for customizable and affordable robotics solutions. As we continue to refine these techniques, the potential for bionic robots to become more integrated into daily life grows, making this an exciting frontier in technology and human-machine collaboration.