The task of laying ceramic tiles remains a labor-intensive and physically demanding process within the construction industry. Workers are required to adopt strenuous kneeling or squatting postures to press tiles firmly into adhesive beds, leading to potential knee injuries and significant physical fatigue over time. Furthermore, the industry faces challenges related to skilled labor shortages and the consistent demand for high-quality, efficient finishing. These factors create a compelling need for automation solutions to improve productivity, ensure consistent quality, and enhance worker safety. The development of robotic systems for tile laying presents a viable and significant solution to these persistent challenges.
Research into robotic tile laying has been ongoing for decades, with systems generally falling into two distinct categories: dedicated tiling machines and modular tiling systems. Dedicated machines are engineered as single-purpose units, often incorporating mobility and a specialized laying mechanism into one integrated chassis. Modular systems, conversely, leverage existing robotic components—such as a mobile base, a general-purpose robotic arm, and a task-specific end effector. This modular approach offers greater flexibility, easier deployment across different job sites, and the potential for reusing core robotic components for other construction tasks. While significant progress has been made, many proposed end effector designs lack comprehensive functionality or detailed, justified engineering analysis. This work addresses that gap by presenting the design, development, and experimental validation of a multi-functional, modular end effector for ceramic tile laying, designed to be mounted on a standard industrial robotic arm.
The primary objective for any tile laying operation, whether manual or automated, is to achieve three key quality criteria: horizontal alignment, where adjacent tiles are flush with consistent grout spacing; vertical alignment, ensuring all tile surfaces lie on a uniform plane; and uniform bedding, where the adhesive forms a complete, void-free bond with the tile back to prevent “hollow” spots or tile detachment. The designed end effector integrates three core functional subsystems to meet these requirements: a vacuum gripping system for pick-and-place, a vibrational compaction system for adhesive settling, and a laser-based metrology system for precise localization. The overall system architecture, comprising a mounting plate, support structure, and integrated components, is illustrated in the figure above. The end effector connects to a robotic arm via a standard interface and receives power and control signals through an integrated umbilical.
Comprehensive Design of the Tile Laying End Effector
1. Vacuum Gripping System for Pick-and-Place
The first critical function of the end effector is to reliably grasp, transport, and release tiles. Given the smooth, non-porous, and rigid surface of glazed ceramic tiles, a vacuum suction system is the optimal choice. This method is non-marring, fast-acting, and can generate substantial holding force with a simple mechanical design. The system’s operating principle is straightforward: evacuating air from the space between a suction cup and the tile surface creates a pressure differential, resulting in a net adhesive force.
The vacuum system for this end effector consists of several key components. A compact, electrically-driven vacuum pump serves as the negative pressure source. A filter protects the pump from dust and adhesive particles. An electrically-controlled solenoid valve manages the connection between the suction circuit and the atmosphere. Finally, a pneumatic manifold distributes the vacuum to an array of individual suction cups. To engage a tile, the vacuum pump is activated while the solenoid valve remains closed, creating suction. To release the tile, the solenoid valve is energized, venting the suction cups to atmospheric pressure and breaking the seal.
The design process involved a careful analysis of operational requirements and component selection:
- Suction Cup Selection: For flat, smooth surfaces like tiles, flat-shaped (bellows-free) suction cups made of nitrile rubber (NBR) are ideal. They offer high sealing efficiency and durability under vertical and horizontal load conditions.
- Vacuum Source Selection: The choice was between a venturi-type vacuum generator (requiring compressed air) and a micro-vacuum pump. For a mobile robotic platform, the micro-vacuum pump is preferred due to its compact size and independence from a large, stationary air compressor.
A critical load analysis was performed to ensure the system could handle standard 300 mm x 300 mm tiles, which typically weigh approximately 1.5 kg. The design uses six suction cups, each with a diameter (d) of 30 mm. The vacuum pump provides a reliable vacuum level (p_v) of 60 kPa (relative negative pressure). A safety factor (t) of 8 is applied to account for dynamic forces, surface imperfections, and the fact that the end effector must operate in both horizontal (floor) and vertical (wall) orientations. The total theoretical holding force (F_total) is calculated as follows:
$$F_{\text{total}} = n \cdot 10^{-3} \cdot \frac{\pi \cdot d^2 \cdot p_v}{4 \cdot t}$$
Substituting the values (n=6, d=0.03 m, p_v=60000 Pa, t=8):
$$F_{\text{total}} = 6 \cdot 10^{-3} \cdot \frac{\pi \cdot (0.03)^2 \cdot 60000}{4 \cdot 8} = 31.8 \, \text{N}$$
The maximum mass (m_max) this force can secure against gravity (g ≈ 9.81 m/s²) is:
$$m_{\text{max}} = \frac{F_{\text{total}}}{g} = \frac{31.8}{9.81} \approx 3.24 \, \text{kg}$$
Since 3.24 kg > 1.5 kg, the design has a substantial safety margin. The key parameters of the vacuum system are summarized below.
| Parameter | Value / Specification |
|---|---|
| Vacuum Level | -60 kPa (Gauge Pressure) |
| Vacuum Source | Electric Micro-Vacuum Pump |
| Suction Cup Diameter | 30 mm |
| Number of Cups | 6 |
| Suction Cup Type | Flat, Nitrile Rubber |
| Calculated Holding Force | 31.8 N |
| Theoretical Max. Payload | > 3.2 kg |
2. Inertial Vibration System for Compaction
After a tile is placed onto the adhesive bed, it is crucial to ensure full and uniform contact to eliminate air pockets, a common defect known as “tiling hollows.” Manual techniques involve tapping the tile with a rubber mallet. The robotic end effector replicates this action through a controlled vibrational compaction system. Applying oscillatory forces helps the tile settle evenly and drives adhesive into microscopic surface irregularities.
The system employs an inertial (or eccentric-rotor) shaker as the actuation mechanism. This design is compact and capable of generating significant dynamic force from a small package. It consists of a DC motor with an unbalanced mass (eccentric) attached to its shaft. As the motor rotates, the centripetal force generated by the eccentric mass creates a sinusoidal inertial force on the motor housing, which is transmitted to the entire end effector structure and, consequently, to the tile. The magnitude of this激振力 (P) is a function of the eccentric mass (m₀), its eccentricity (r), and the rotational speed (ω):
$$P = m_0 r \omega^2$$
Assuming the end effector and tile move as a near-rigid body in the vertical direction during low-frequency compaction, the approximate vertical force (F_y) transmitted is:
$$F_y \approx m_0 r \omega^2 \sin(\omega t)$$
The component selection was driven by the need for a controlled, adjustable force within the payload limits of common collaborative robots. A small brushed DC motor (RS545 type) was chosen for its good speed-torque characteristics and ease of control via voltage. An eccentric mass was designed from steel, with a mass m₀ = 0.08 kg and an effective eccentricity r = 0.0084 m. To achieve a target force amplitude of approximately 50 N, the required angular velocity is derived from:
$$F_{\text{target}} = m_0 r \omega^2$$
$$\omega = \sqrt{\frac{F_{\text{target}}}{m_0 r}} = \sqrt{\frac{50}{0.08 \times 0.0084}} \approx 273 \, \text{rad/s}$$
This corresponds to a rotational speed of:
$$n = \frac{\omega \cdot 60}{2\pi} \approx 2600 \, \text{RPM}$$
By regulating the input voltage to the DC motor, the rotational speed—and thus the compaction force—can be precisely tuned for different tile sizes, adhesive types, or substrate conditions.
3. Laser Metrology System for Localization
Precise tile placement is paramount for achieving perfect alignment. The end effector is equipped with a metrology system to locate the tile’s position and orientation relative to the robot and previously laid tiles. The system utilizes four laser distance sensors mounted orthogonally around the perimeter of the end effector structure. These sensors emit a collimated beam and measure the time-of-flight or phase shift of the reflected light to determine distance with high accuracy.
The localization algorithm is based on an edge-finding routine. The robotic arm maneuvers the end effector so that the laser spots from two perpendicular sensors scan across the expected edges of a reference tile or a previously laid tile. As a sensor crosses from the substrate onto the tile (or vice-versa), the measured distance exhibits a sharp discontinuity. The robot’s controller records its precise Cartesian position at the moment this edge is detected by each sensor. By collecting data points from two perpendicular edges, the system can construct a mathematical model of the tile’s pose. Assuming the tile is a perfect rectangle, the intersection point of the two defined edge lines provides a precise corner location. Combining this with the known tile dimensions allows the system to calculate the exact position and orientation required for placing the next tile with the correct grout spacing. This closed-loop feedback corrects for any cumulative errors in robot positioning or adhesive bed variability.
System Integration and Software Architecture
To validate the end effector design, a complete robotic tile-laying cell was developed. The hardware platform consists of a 6-degree-of-freedom (6-DOF) collaborative robotic arm (Aubo-i10 model), the custom-designed end effector mounted on its flange, the external vacuum pump module, and a central control computer. The Robot Operating System (ROS) framework, specifically the Noetic distribution on Ubuntu 20.04, was chosen as the software backbone due to its flexibility, strong community support, and rich ecosystem of tools for robotics.
The system integration involved several key steps:
- Unified Robot Description (URDF): A detailed 3D model of the robotic arm and the end effector was created and converted into a URDF file. This file defines the kinematic chain, joint limits, collision geometries, and visual meshes, forming the essential digital twin of the physical system.
- MoveIt! Configuration: The MoveIt! motion planning framework was configured using the generated URDF. This setup enables advanced capabilities such as inverse kinematics solving (using the TRAC-IK solver), collision-aware path planning (using the RRT-Connect algorithm), and trajectory execution. It allows the system to plan safe, efficient paths for picking tiles from a stack and placing them at the target location.
- Custom ROS Nodes: Dedicated software nodes were developed to manage the end effector‘s specific functions:
- A hardware interface node communicates with the robot controller’s I/O to toggle the vacuum solenoid valve, enable/disable the vibration motor, and read analog signals from the laser sensors.
- A localization node implements the edge-search algorithm, processes laser data, and computes the tile’s pose.
- A task sequencer node orchestrates the overall workflow: initiating a pick operation, moving to the laying area, performing the edge-finding routine for precise positioning, executing the place and vibrate commands, and then returning for the next tile.
This modular software architecture ensures that the end effector functions are neatly encapsulated and can be controlled through high-level commands, making the system adaptable and easier to debug or extend.
Experimental Validation and Performance Analysis
1. Repeatable Localization Accuracy Test
The accuracy of the laser-based localization system is fundamental to the overall placement precision. An experiment was conducted to evaluate its repeatability. A single 300 mm x 300 mm test tile was fixed in place. The robotic arm, equipped with the end effector, was programmed to perform the edge-finding routine 20 consecutive times on the same tile. For each run, the coordinates of the four edge points were recorded via the laser sensors and the robot’s forward kinematics. The calculated center position of the tile from each trial was then compared.
The raw distance readings from the laser sensors showed a maximum standard deviation of 0.238 mm, indicating very low measurement noise. The derived tile center positions in the X, Y, and Z axes exhibited random error distributions. The standard deviation of the tile’s calculated position was determined to be less than 0.343 mm in all Cartesian directions. This level of repeatability is more than sufficient for tile laying, where typical grout lines are several millimeters wide, confirming the metrology system’s capability to provide highly reliable pose estimation.
2. Pick-and-Place and Laying Functionality Test
The core functionality of the vacuum gripping system was tested under simulated real-world conditions. A batch of 300 mm x 300 mm tiles (mass ≈ 1.45 kg) was used. The end effector performed 50 pick-and-place cycles in two critical orientations: horizontal (simulating floor tiling) and vertical (simulating wall tiling). After successful adhesion, the robot executed a linear translation of 30 cm at a speed of 10 cm/s before releasing the tile. This test evaluated the grip security under dynamic motion.
Subsequently, a continuous laying sequence of six tiles was performed on a prepared adhesive bed to demonstrate the complete integrated workflow, including placement and vibrational compaction.
The results were highly positive. The vacuum system successfully handled all 50 test cycles in both orientations with a 100% success rate, demonstrating robust adhesion and safety margin. The full laying sequence proved that the end effector could reliably execute the entire cycle—pick, transport, locate, place, and compact. The table below summarizes the key test conditions and outcomes.
| Test Aspect | Conditions & Results |
|---|---|
| Tile Specs | 300×300 mm, ~1.45 kg |
| Tested Orientations | Horizontal & Vertical |
| Motion Profile | 30 cm translation @ 10 cm/s |
| Cycles per Orientation | 50 |
| Gripping Success Rate | 100% |
| Full Laying Demo | 6 tiles laid sequentially |
| Key Achievement | Stable operation in all required poses |
Conclusion and Future Perspectives
This work has presented the complete design and validation of a sophisticated, modular end effector for robotic ceramic tile laying. The end effector successfully integrates three essential functions into a single, coherent package: a vacuum-based gripping system for reliable pick-and-place, an inertial vibration system for adhesive compaction and void elimination, and a laser-based metrology system for high-precision tile localization. The modular nature of this end effector allows it to be mounted on various industrial or collaborative robotic arms, which can in turn be deployed on AGVs, rails, or stationary frames to suit different construction site layouts.
Experimental validation confirmed the system’s capabilities. The laser localization system achieved a repeatability standard deviation better than 0.343 mm, ensuring precise alignment. The vacuum gripper demonstrated 100% reliability in handling tiles in both horizontal and vertical orientations. The integrated system successfully executed complete tile laying sequences, proving the practical viability of the design.
The developed end effector provides a solid foundation for automating a key construction finishing task. Future work will focus on several enhancements: integrating a vision system for full pose estimation of arbitrarily stacked tiles, developing more advanced compaction strategies informed by force/torque sensing, and creating higher-level planning software for optimizing tile laying patterns in complex room layouts. By continuing to refine this modular end effector approach, the path toward widespread adoption of robotic automation in construction becomes increasingly clear, promising significant benefits in efficiency, quality, and worker well-being.