In the context of global efforts toward carbon neutrality, the advancement of clean and efficient power generation technologies is paramount. Solar photovoltaic power generation, as a renewable energy source, offers significant advantages in safety, efficiency, economy, and environmental friendliness. However, solar panels are often installed in outdoor environments where they are susceptible to accumulation of dust and bird droppings. These contaminants reduce light transmittance, decrease power generation efficiency, and can lead to hotspot formation, damaging the panel structure over time. Traditional cleaning methods, such as manual washing, are inefficient and hazardous, especially in challenging terrains like wetlands and hilly areas where solar panels are mounted at heights with limited access. To address this, I have developed an autonomous cleaning robot based on vision-photoelectric sensors, designed to operate on tilted solar panels in complex environments. This article details the design, analysis, path planning, cleaning strategies, and experimental validation of the robot, emphasizing its ability to enhance the performance and longevity of solar panels.
The robot is engineered to navigate and clean solar panels autonomously, leveraging a fusion of visual and photoelectric sensors for adaptive path planning and targeted cleaning. Its compact and flexible design allows it to be deployed via a tracked lifting platform onto solar panels, eliminating the need for additional infrastructure like rails. The core components include a chassis, side brushes, a roller brush, misting nozzles, tracked wheels, photoelectric sensors, vacuum generators, a main control box, a vision sensor, and a water tank. The vision sensor captures surface features of the solar panels, while photoelectric sensors detect edges to prevent falls. The vacuum generators provide suction to adhere the robot to the glass surface, enabling operation on tilted panels. During cleaning, water is sprayed onto the panel, and the brushes remove contaminants. The control system integrates sensor data to execute movements and cleaning actions, ensuring comprehensive coverage and efficient operation.
To ensure stable operation on tilted solar panels, the robot’s adhesion conditions and required suction force were analyzed. Solar panels are typically installed at an optimal angle to maximize sunlight exposure; in many regions, this angle is around 20 degrees. The robot must maintain traction without slipping or falling. The driving force is provided by motors that rotate the tracked wheels, generating friction between the tracks and the panel surface. The frictional force has a limit defined as the adhesion force, which depends on the normal force and the coefficient of friction. For the tracks to avoid slipping, the tangential force must not exceed the adhesion force. This can be expressed as:
$$F_s = \mu F_n$$
where \(F_s\) is the adhesion force, \(\mu\) is the adhesion coefficient between the track and the panel surface, and \(F_n\) is the normal force. The condition for no slipping is:
$$\frac{F_x}{F_n} = C \leq \mu$$
where \(F_x\) is the tangential force from the driving torque, and \(C\) is the adhesion rate. The robot’s weight and suction force contribute to the normal force. Considering a robot mass of 8 kg (including water) and a panel tilt angle \(\theta = 20^\circ\), the normal force is:
$$F_N = mg \cos \theta + F_r$$
where \(m\) is the total mass, \(g\) is gravitational acceleration, and \(F_r\) is the suction force from the vacuum generators. To prevent sliding, the frictional force must satisfy:
$$F_S = mg \sin \theta < f_s F_N$$
where \(f_s\) is the static friction coefficient. On wet solar panels, the adhesion coefficient can drop to 0.25, making suction essential. Solving for the minimum suction force, we find \(F_r > 33.62\, \text{N}\). Using vacuum generators with bell-shaped suction cups of diameter 50 mm and a safety factor of 4, the required vacuum pressure is calculated as:
$$F_r = \frac{\pi D^2 p}{4n} \Rightarrow p = \frac{4n F_r}{\pi D^2}$$
Substituting values yields \(p \approx 68.49\, \text{kPa}\). Standard vacuum generators with a maximum vacuum of 88 kPa are sufficient. Two vacuum generators are used to mitigate pressure loss when crossing gaps between solar panels.
Path planning is critical for autonomous navigation on solar panels. The robot uses a vision sensor and photoelectric sensors to perceive its environment and adjust its trajectory. Initially, the robot may be placed arbitrarily on the panel, so it must first orient itself toward the nearest edge. The vision sensor captures images of the panel, which are processed to detect aluminum frames around the edges. Image processing steps include Gaussian filtering for noise reduction, grayscale conversion, edge detection using the Roberts operator, thresholding for segmentation, and area filtering to remove noise. Hough line detection is then applied to fit lines to the edges, and column scanning computes the pixel distance to the nearest frame. Based on this, the robot rotates to face the closest edge and moves forward until photoelectric sensors detect the frame.
The photoelectric sensors at the front and corners provide real-time edge detection. When the robot approaches an edge, sensors emit low-level signals, prompting the control system to stop or adjust direction. For example, if the left-front sensor detects an edge, the robot rotates clockwise until the right-front sensor also detects it, aligning the robot parallel to the frame. By combining sensor signals and vision data, the robot can identify its corner position on the panel. A decision table is used to determine the corner based on sensor inputs and vision sensor rotation angles. Once the robot reaches a corner, it follows a predefined path to the starting position for cleaning operations.
| Vision Sensor Angle (°) | Left-Rear Signal | Right-Rear Signal | Corner Position |
|---|---|---|---|
| 90 or 270 | Present | Absent | Top-Left |
| 180 | Present | Absent | Top-Right |
| 90 or 270 | Absent | Present | Top-Right |
| 180 | Absent | Present | Top-Left |
Cleaning strategies are tailored for different contaminants: dust and bird droppings. For dust, which is uniformly distributed, the robot employs a boustrophedon (back-and-forth) pattern to cover the entire panel surface. This path minimizes turns and adapts to panel dimensions, with the robot moving horizontally across the panel while accounting for gaps between panels. Photoelectric sensors ensure the robot stops before edges to prevent falls. For bird droppings, which are localized, the robot uses image recognition to identify and target them. After completing dust cleaning, the robot captures an image of the panel and processes it to detect bird droppings. The image is filtered, thresholded, and analyzed to locate contaminants relative to panel features like grid lines. The center of the bird dropping is calculated, and its distance from the nearest grid line is determined using pixel measurements and actual panel dimensions. Based on this, the robot moves to the precise location for repeated cleaning.
The positioning of bird droppings involves geometric calculations. Let the pixel distances from the contaminant center to the left and right frames be \(l_M\) and \(l_N\), and the pixel lengths of the nearest grid lines be \(l_P\) and \(l_Q\). The actual distances \(l_1\) and \(l_2\) are found using similarity ratios:
$$\frac{l_M}{l_1} = \frac{l_Q}{l_4}, \quad \frac{l_N}{l_2} = \frac{l_P}{l_3}$$
where \(l_3\) and \(l_4\) are the actual lengths of the grid lines. Solving for the actual distances:
$$l_1 = \frac{l_M l_4}{l_Q}, \quad l_2 = \frac{l_N l_3}{l_P}$$
These distances guide the robot to the target with high accuracy.
Experiments were conducted to evaluate the robot’s performance. A testbed was built using tempered glass to simulate solar panels, with dimensions of 1 m × 1 m and aluminum frames. The panel was tilted at 20 degrees, and contaminants like fine soil and small clumps were spread to mimic dust and bird droppings. The robot was deployed randomly on the panel, and its ability to reach the starting position, clean dust, and locate bird droppings was measured. Cleaning efficiency was assessed by timing the operations, and cleaning quality was evaluated through image analysis to compute dust removal rates. The robot’s positioning error for bird droppings was also calculated.
| Trial | Time to Initial Position (s) | Time for Full Cleaning (s) | Positioning Error (mm) |
|---|---|---|---|
| 1 | 49.3 | 72.6 | 1.5 |
| 2 | 47.6 | 70.3 | 1.2 |
| 3 | 53.1 | 68.8 | 1.4 |
| 4 | 51.8 | 74.4 | 1.3 |
| 5 | 46.7 | 73.9 | 1.5 |
The average time to reach the initial position was 49.7 seconds, and the average time for full cleaning was 72.0 seconds. This translates to a cleaning efficiency of approximately 50 m²/h. The average positioning error for bird droppings was 1.38 mm, demonstrating high accuracy. To assess cleaning quality, images of the solar panels were analyzed before and after cleaning. The images were grayscaled, filtered with median filtering, and thresholded to segment dust particles. The number of white pixels representing dust was counted, and the dust removal rate was computed as the reduction in dust pixels relative to the initial count. Results showed an average dust removal rate of 91.16%, indicating effective cleaning.
| Metric | Average Value |
|---|---|
| White pixel count (clean panel) | 26122 |
| White pixel count (before cleaning) | 35677 |
| White pixel count (after cleaning) | 26967 |
| Dust removal rate | 91.16% |
In conclusion, the vision-photoelectric sensor-based cleaning robot offers a robust solution for maintaining solar panels in challenging environments. By analyzing adhesion conditions and suction requirements, the robot can operate safely on tilted surfaces. The integration of visual and photoelectric sensors enables adaptive path planning and precise targeting of contaminants. Experimental results confirm high cleaning efficiency, accurate positioning, and effective dust removal. This technology contributes to the optimization of solar energy generation by ensuring that solar panels remain clean and efficient, supporting the transition to renewable energy. Future work may focus on enhancing autonomy, such as incorporating machine learning for better contaminant classification, or improving power efficiency for longer operation cycles. Overall, this robotic system represents a significant step toward automated maintenance of solar panels, reducing labor costs and increasing the reliability of solar power installations.
The development of this robot underscores the importance of innovation in renewable energy maintenance. As solar panels become more widespread, automated cleaning solutions will play a crucial role in maximizing their output and lifespan. By addressing the specific challenges of dust and bird droppings, this robot demonstrates how sensor fusion and intelligent control can be applied to real-world problems. The use of solar panels in diverse environments, from deserts to wetlands, requires adaptable technologies, and this robot’s design offers a flexible approach. Continued research and testing will further refine its capabilities, potentially integrating with smart grid systems for proactive maintenance. Ultimately, the goal is to create sustainable and efficient energy systems, where solar panels operate at peak performance with minimal human intervention.