Flexible perovskite robot based on environmental perception and autonomous phototactic charging and control method thereof

Abstract

The invention discloses a flexible perovskite robot based on environmental perception and autonomous phototactic charging and a control method thereof, and belongs to the technical field of intelligent robots, the control method comprises the following steps: S1, collecting environmental perception data and robot state data; s2, generating an illumination semantic map based on the environmental perception data through an SLAM technology and a Gaussian process regression algorithm; s3, when the electric quantity is lower than a threshold value, the task is interrupted, and based on the illumination semantic map and the robot state data, an optimal charging strategy is output through a multi-objective optimization algorithm; s4, the robot navigates to a target charging position according to the optimal charging strategy; and S5, attitude dynamic optimization is started, autonomous charging is executed, and when the electric quantity reaches a threshold value, an interrupt task is restarted, so that the robot can autonomously search for the optimal illumination position and adjust the attitude to maximize the energy collection efficiency through deep integration of environment perception, intelligent decision and dynamic execution.

Description

Technical Field

[0001] This invention belongs to the technical field of intelligent robots, specifically, it relates to a flexible perovskite robot based on environmental perception and autonomous phototactic charging, and its control method. Background Technology

[0002] With the increasing popularity of companion robots, their energy supply issues are becoming more and more prominent. Current technologies mainly rely on fixed charging stations for energy replenishment, which requires interrupting tasks to return to charging, seriously affecting the continuity of robot work and user experience.

[0003] Although existing research has attempted to apply solar energy technology to robotics, the following limitations exist:

[0004] (1) Most of them use rigid solar panels, which cannot adapt to the curved shape of the robot;

[0005] (2) It can only achieve static charging and cannot effectively collect energy while moving;

[0006] (3) Lacking intelligent energy management strategies, it cannot autonomously optimize charging behavior based on ambient light conditions.

[0007] In recent years, flexible perovskite solar cells have attracted attention due to their excellent low-light performance, flexibility, and high conversion efficiency. However, a mature solution for their application in the autonomous charging of mobile robots is still lacking. In particular, there are still technological gaps in the systematic innovation of combining ambient light perception, autonomous movement decision-making, and dynamic attitude adjustment. Summary of the Invention

[0008] In view of this, in order to solve the above-mentioned problems of the prior art, the purpose of the present invention is to provide a flexible perovskite robot based on environmental perception and autonomous phototactic charging and its control method, so as to achieve the goal of enabling the robot to autonomously find the best lighting position and adjust its posture to maximize energy harvesting efficiency through deep integration of environmental perception, intelligent decision-making and dynamic execution.

[0009] The technical solution adopted in this invention is: a flexible perovskite robot based on environmental perception and autonomous phototactic charging, the robot comprising:

[0010] A multimodal environment perception module is used to monitor ambient lighting information, three-dimensional environmental information, and robot posture status information.

[0011] An ambient lighting map construction module generates a lighting map based on information collected by a multimodal environment perception module.

[0012] An autonomous charging decision module outputs the optimal charging strategy based on a lighting map.

[0013] A dynamic attitude optimization module, which performs charging according to the optimal charging strategy and dynamically adjusts its own attitude in real time during the charging process;

[0014] A multimodal energy management module is connected to a flexible perovskite power generation module, a lithium polymer battery pack, and a supercapacitor module. The flexible perovskite power generation module covers the surface of the robot's outer shell.

[0015] The main controller is communicatively connected to the multimodal environment perception module, the ambient lighting map construction module, the autonomous charging decision module, the dynamic attitude optimization module, and the multimodal energy management module.

[0016] Furthermore, the multimodal environment perception module includes a distributed illumination sensor array, a 3D vision unit, and an inertial measurement unit.

[0017] Furthermore, the flexible perovskite power generation module includes several flexible perovskite battery units, each of which is connected to an MPPT controller, and each of the flexible perovskite battery units is distributed in different parts of the robot's body.

[0018] Furthermore, the robot also includes: a robot payload, a robot walking mechanism, and a robot servo mechanism. The robot payload, robot walking mechanism, and robot servo mechanism are electrically connected to the multimodal energy management module, and the robot walking mechanism is communicatively connected to the autonomous charging decision module, while the robot servo mechanism is communicatively connected to the dynamic posture optimization module.

[0019] This invention also provides a control method for a flexible perovskite robot based on environmental perception and autonomous phototactic charging, the control method comprising:

[0020] S1: Collect environmental perception data and robot status data;

[0021] S2: Based on environmental perception data, a semantic map of illumination is generated using SLAM technology and Gaussian process regression algorithm;

[0022] S3: When the battery level is below the threshold, the task is interrupted and the optimal charging strategy is output through a multi-objective optimization algorithm based on the illumination semantic map and robot state data.

[0023] S4: The robot navigates to the target charging location according to the optimal charging strategy;

[0024] S5: Initiate dynamic attitude optimization and perform autonomous charging. When the battery level reaches the threshold, restart the interrupt task.

[0025] Furthermore, the control method also includes:

[0026] S6: Dynamically monitors charging power during the charging process and automatically switches between different power supply modes based on the charging power through the energy management unit.

[0027] Furthermore, the environmental perception data includes: ambient lighting data, 3D vision data, and robot posture data;

[0028] The robot status data includes: robot battery status, task priority, robot energy consumption cost, robot charging efficiency, and user behavior patterns.

[0029] Furthermore, the generation logic of the illumination semantic map is as follows:

[0030] S201: Constructing an environmental spatial structure map based on SLAM technology;

[0031] S202: Generating a light intensity distribution map using a Gaussian process regression algorithm;

[0032] S203: Integrate the light intensity distribution map into the time dimension and predict the light time variation pattern based on the Gaussian process regression learning model;

[0033] S204: Integrate environmental spatial structure map, light intensity distribution map, and light time variation pattern with semantic information to generate a light semantic map.

[0034] Furthermore, the output logic of the optimal charging strategy is as follows:

[0035] S301: Acquire robot state data and lighting semantic map;

[0036] S302: Output the optimal charging strategy through a multi-objective optimization algorithm, and the optimal charging strategy includes: target charging location, optimal movement path and estimated charging time.

[0037] Furthermore, the logic for the dynamic attitude optimization is as follows:

[0038] S501: Adjust the robot's orientation according to the direction of the light;

[0039] S502: Adjust the angle of the robot's head and body;

[0040] S503: Sample the current power generation and determine whether the power generation has increased. If yes, continue to correct at a fixed frequency in the current adjustment direction and execute S502; if no, return to S502 and adjust in the opposite direction.

[0041] S504: Determine whether the power generation has reached the optimal threshold. If yes, end the adjustment and maintain the current attitude; otherwise, return to S503.

[0042] The beneficial effects of this invention are as follows:

[0043] 1. The flexible perovskite robot and its control method based on environmental perception and autonomous phototactic charging provided by this invention can autonomously find a light-illuminated area and perform autonomous charging when the robot's power is lower than a preset threshold. After charging reaches the preset threshold, the robot continues to perform its tasks. It can work 24 hours a day in an indoor environment and truly achieve energy self-management. Attached Figure Description

[0044] Figure 1 This is a system architecture diagram of the flexible perovskite robot based on environmental perception and autonomous phototactic charging provided by the present invention.

[0045] Figure 2 This is a logic diagram for constructing a semantic map of illumination in the flexible perovskite robot control method based on environmental perception and autonomous phototactic charging provided by the present invention.

[0046] Figure 3 This is the logic diagram of the output optimal charging strategy in the flexible perovskite robot control method based on environmental perception and autonomous phototactic charging provided by the present invention.

[0047] Figure 4 This is the logic diagram of dynamic posture optimization in the flexible perovskite robot control method based on environmental perception and autonomous phototactic charging provided by the present invention.

[0048] Figure 5 This is a schematic diagram of the multimodal energy management logic in the flexible perovskite robot control method based on environmental perception and autonomous phototactic charging provided by the present invention. Detailed Implementation

[0049] The embodiments of this application are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar modules or modules having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain this application, and should not be construed as limiting this application. Rather, the embodiments of this application include all variations, modifications, and equivalents falling within the spirit and scope of the appended claims.

[0050] Example 1

[0051] like Figure 1As shown, this embodiment provides a flexible perovskite robot based on environmental perception and autonomous phototactic charging. It utilizes flexible perovskite power generation technology, an ambient light perception system, an autonomous mobile charging decision algorithm, and a dynamic posture optimization-based companion robot energy autonomy system. It is suitable for long-term autonomous work scenarios such as home service, elderly care, and child companionship. Specifically, the robot includes the following parts:

[0052] ① Main controller

[0053] The main controller is connected to the multimodal environment perception module, the ambient lighting map construction module, the autonomous charging decision module, the dynamic attitude optimization module, and the multimodal energy management module. After collecting signals from each module, the main controller runs the corresponding algorithm logic and issues the corresponding execution instructions.

[0054] ② Multimodal environment perception module

[0055] This multimodal environmental perception module monitors ambient lighting information, 3D environmental information, and robot posture information. Specifically, ambient lighting information is collected through a distributed lighting sensor array, with eight multispectral lighting sensors deployed on the robot's head, chest, and back to monitor ambient light intensity, spectral composition, and incident angle in real time. 3D environmental information is acquired through a 3D vision system, specifically using an RGB-D camera. Robot posture information is acquired through an inertial measurement unit (IMU) to monitor the robot's posture and motion status in real time. The distributed lighting sensor array, 3D vision system, and IMU are all communicatively connected to the main controller. The lighting sensors are AMS AS7341 multispectral sensors, connected to the main controller via an I2C bus, with a sampling frequency of 10Hz. The 3D vision system uses an Intel RealSense D435i camera, providing 640x480 resolution depth information.

[0056] For example, when the robot is patrolling the living room, the head sensor detects that the light intensity in the direction of the window is 350 lux. Spectral analysis shows that it is mainly scattered light. The IMU records that the current angle between the robot's orientation and the direction of the light is 45 degrees.

[0057] ③Ambient lighting map construction module

[0058] The ambient lighting map construction module generates a lighting map based on information collected by the multimodal environment perception module. This module uses SLAM technology to construct an environmental spatial structure map containing lighting information, predicts the light intensity distribution at different locations and times using a Gaussian process regression algorithm, and generates a lighting semantic map through data fusion. The lighting semantic map contains the following hierarchical information: ① static obstacle layer; ② light intensity distribution layer; ③ lighting time variation pattern layer; ④ optimal charging location recommendation layer.

[0059] For example, through model learning and training, the light intensity in the area near the living room sofa is consistently above 400 lux between 9 and 11 am, and no one passes by frequently, so this area is marked as the best charging location.

[0060] ④ Autonomous charging decision module

[0061] The autonomous charging decision module outputs the optimal charging strategy based on the illumination map. When the autonomous charging decision module is running, the autonomous charging decision model comprehensively considers the following factors: 1) current state of charge (SOC); 2) task priority; 3) energy cost of moving to the candidate charging point; 4) predicted charging efficiency; and 5) user behavior pattern.

[0062] The final output of the optimal charging strategy includes: target location, movement path, and estimated charging time. The target location is the best charging location, and the movement path is the walking path to avoid obstacles.

[0063] For example, when the robot's battery level drops to 30%, the autonomous charging decision module assesses the current task as a low-priority patrol, calculates the net energy gain of moving to the balcony charging point as +8500J, and takes 120 seconds, then initiates the autonomous charging process.

[0064] ⑤ Dynamic Attitude Optimization Module

[0065] The dynamic attitude optimization module performs charging according to the optimal charging strategy and dynamically adjusts its own attitude in real time during the charging process; wherein, attitude adjustment includes:

[0066] 1) Coarse adjustment: The robot's chassis is moved to orient the entire robot towards the optimal lighting direction;

[0067] 2) Fine adjustment: Adjust the angle of the robot's upper body and head by using the robot's servo motors to keep the power generation surface perpendicular to the direction of light incidence.

[0068] For example, when the robot is charging on the balcony, it first turns its body towards the window, then continuously fine-tunes the angle of its upper body, samples the power generation every 2 seconds, and adjusts in the opposite direction when it detects a decrease in power, eventually finding the posture with the highest power generation efficiency (e.g., tilt angle of 32° and deflection angle of 15°).

[0069] ⑥ Multimodal Energy Management Module

[0070] The multimodal energy management module connects a flexible perovskite power generation module, a lithium polymer battery pack, and a supercapacitor module. The module employs a dynamic power allocation strategy, prioritizing the use of the flexible perovskite power generation module for power supply, with excess energy stored in the lithium polymer battery pack; when instantaneous high power is required or solar power generation is insufficient, a hybrid power supply is used.

[0071] The flexible perovskite power generation module covers the robot's outer shell surface. Specifically, the flexible perovskite power generation module includes several flexible perovskite battery cells, each connected to an MPPT controller. Each flexible perovskite battery cell is independently equipped with a micro MPPT controller, which can optimize power generation efficiency in real time. These flexible perovskite battery cells are distributed in different parts of the robot's body, such as the head, back, sides of the torso, and outer sides of the limbs. The flexible perovskite battery cells utilize a perovskite active layer optimized with guanidine sulfamate (GAS) doping at a concentration of 0.8-1.5 mg / ml, achieving a photoelectric conversion efficiency of 18-22% under indoor lighting conditions of 200-500 lux. The flexible perovskite battery cells are modularly designed, covering more than 85% of the robot's outer shell surface, including key areas such as the curved surface of the head, sides of the torso, and outer sides of the limbs. For example, the main power plate on the back of the robot measures 200mm × 150mm and adopts a hexagonal honeycomb structure design, consisting of 24 independent perovskite cell units. Each perovskite cell unit is equipped with an individual MPPT controller. Under 300 lux illumination, a single unit can output 0.5W of power, and the maximum output power of the entire back plate is 12W.

[0072] The lithium polymer battery pack uses a high-energy-density lithium polymer battery pack with a rated capacity of 6000mAh; the supercapacitor module has a capacity of 100F and is used for peak power buffering.

[0073] The robot also includes a robot payload, a robot locomotion mechanism, and a robot servo mechanism, all of which are electrically connected to the multimodal energy management module. The robot locomotion mechanism is communicatively connected to the autonomous charging decision module to move the robot to the optimal charging position according to the output optimal charging strategy. The robot servo mechanism is communicatively connected to the dynamic posture optimization module to adjust the robot's orientation and body angle to the optimal posture according to posture optimization commands.

[0074] The workflow of the robot during operation is as follows:

[0075] The robot continuously monitors ambient light and its own status. When it needs to be charged, it selects the optimal charging location based on the light semantic map, autonomously navigates to the target location, and adjusts its posture dynamically to maximize power generation efficiency. At the same time, it can perform low-priority tasks during the charging process, and returns to normal working mode after the battery is fully charged.

[0076] For example, at 3 p.m., the robot's battery dropped to 40%. It detected that the light intensity in the study desk area reached 280 lux and no one was using it. So it moved to the area, adjusted its posture so that the power panel on its back faced the window, and performed environmental monitoring tasks while charging. Two hours later, the battery was restored to 85% and it returned to normal working mode.

[0077] Example 2

[0078] like Figures 2-5 As shown, the present invention also provides a control method for a flexible perovskite robot based on environmental perception and autonomous phototactic charging, the control method comprising:

[0079] S1: Collect environmental perception data and robot status data; specifically, environmental perception data includes: ambient lighting data collected by the light sensor, 3D vision data collected by the 3D vision system, and robot posture data obtained by the inertial measurement unit; on the other hand, robot status data is collected by feedback from various functional modules inside the robot, including: robot power status (SoC), task priority, robot energy consumption cost (e.g., power consumption for walking), robot charging efficiency, and user behavior patterns.

[0080] S2: Based on environmental perception data, a lighting semantic map is generated using SLAM technology and Gaussian process regression algorithm; wherein, the generation logic of the lighting semantic map is as follows:

[0081] S201: Construct an environmental spatial structure map based on SLAM technology. The SLAM model relies on environmental information collected by various sensors. Based on the environmental features perceived by the sensors and the localization results, the map is gradually constructed. This environmental spatial structure map provides the robot with a navigation path and helps it avoid obstacles marked on the map.

[0082] S202: Generate a light intensity distribution map using the Gaussian process regression algorithm; specifically, for the environmental spatial structure map constructed by SLAM, use the trained Gaussian process regression algorithm (GPR model) to predict the light intensity, overlay the prediction results with the spatial map and use color to distinguish the light intensity to generate a light intensity distribution map.

[0083] S203: Integrating light intensity distribution maps into the time dimension, and predicting the temporal variation of light intensity based on a Gaussian process regression learning model; it includes:

[0084] S2031: An environmental spatial structure map based on SLAM is constructed to determine the coordinates of key locations in the environment. For each location, illumination data is repeatedly collected at different time points to obtain training samples containing "location + time + illumination intensity".

[0085] S2032: Train a Gaussian process regression learning model using the collected "location + time + light intensity" samples;

[0086] S2033: Predict the pattern of light duration variation by using a trained Gaussian process regression learning model. For example: tomorrow at 10:00, the light at location A will be 500 lux (suitable for charging), while the light at location B will be only 100 lux.

[0087] S204: Integrate environmental spatial structure map, light intensity distribution map, and light time variation pattern with semantic information to generate a light semantic map. For example, use a knowledge graph and deep learning fusion model for deep integration to generate a light semantic map with semantic labels. This light semantic map includes: "path planning + light intensity features + time pattern + optimal charging label".

[0088] S3: When the robot's battery level falls below a threshold, the current task is automatically interrupted, and based on the illumination semantic map and robot state data, a multi-objective optimization algorithm is used to output the optimal charging strategy. The output logic of the optimal charging strategy is as follows:

[0089] S301: Acquire robot status data and illumination semantic map; As mentioned above, robot status data includes: robot power status (SoC), task priority, robot energy consumption cost (e.g., walking power consumption), robot charging efficiency, and user behavior pattern; Meanwhile, the illumination semantic map includes: "path planning + illumination intensity features + time pattern + optimal charging label".

[0090] S302: Outputs the optimal charging strategy through a multi-objective optimization algorithm, whereby the optimal charging strategy includes: target charging location, optimal movement path, and estimated charging time. Specifically, the multi-objective optimization algorithm executes every 30 seconds, comprehensively evaluating information such as battery level, task priority, and light intensity to update the charging strategy. This multi-objective optimization algorithm is trained based on the Q-learning reinforcement learning framework, and the reward function includes multiple dimensions such as energy gain, task completion rate, and user satisfaction. The logic of the multi-objective optimization algorithm includes:

[0091] S3021: Update the robot status and lighting semantic map status every 30 seconds;

[0092] S3022: Set the reward function (R) = Energy Gain Score × a + Task Completion Score × b + User Satisfaction Score × c, where a, b, and c are weights, respectively; the Energy Gain Score corresponds to charging efficiency and energy consumption cost; the Task Completion Score corresponds to task priority and battery status; and the User Satisfaction Score corresponds to user behavior pattern.

[0093] S3023: Based on the Q-learning reinforcement learning framework, a "Q-table" is generated, and the action with the highest reward score in the "Q-table" is used as the optimal charging strategy;

[0094] S3024: After execution, the "Q table" is modified based on the actual reward score, thereby dynamically updating the "Q table" and optimizing the Q-learning reinforcement learning framework.

[0095] S4: The robot navigates to the target charging location according to the optimal charging strategy. The robot calls the mobile chassis (walking mechanism) through the main controller to navigate to the target charging location according to the planned optimal path.

[0096] S5: Initiate dynamic attitude optimization and perform autonomous charging. When the battery level reaches the threshold, restart the currently interrupted task. The logic for dynamic attitude optimization is as follows:

[0097] S501: Adjust the robot's orientation according to the direction of light, that is, the side covered with the flexible perovskite power generation unit faces the direction of light; the orientation and posture adjustment is achieved through the chassis, which can rotate 360° without restriction.

[0098] S502: Adjust the robot's head and body angles to align the largest area of ​​the flexible perovskite power generation unit perpendicular to the direction of illumination. The robot's head and body surfaces are covered with flexible perovskite power generation units, which use a flexible substrate with a thickness of 0.3 mm. The perovskite active layer is prepared using a slot coating process, with a GAS doping concentration of 1.2 mg / ml, and can withstand repeated bending with a maximum bending radius of 3 mm. The adjustment of the head and body angles is achieved through servo motors, with a pitch angle range of ±45° for the upper body and a head rotation angle range of ±60°.

[0099] S503: Sample the current power generation and determine whether the power generation has increased. If yes, continue to correct at a fixed frequency according to the current adjustment direction and execute S502. A PID control algorithm can be used, and the attitude is adjusted once every 2 seconds. The power generation efficiency is used as the feedback signal. If no, it means that the current angle adjustment direction is wrong. If necessary, return to S502 and adjust in the opposite direction.

[0100] S504: Determine whether the power generation has reached the optimal threshold. If yes, end the adjustment and maintain the current attitude; otherwise, return to S503.

[0101] S6: During charging, the charging power is dynamically monitored, and different power supply modes are automatically switched through the energy management unit based on the charging power. Specifically, the energy management unit uses Texas Instruments' BQ25723 multi-chemistry battery charging manager, which supports maximum power point tracking for solar input, has a charge / discharge efficiency of over 92%, and supports three operating modes: pure solar power mode, hybrid power mode, and pure battery mode, automatically switching according to lighting conditions and load requirements.

[0102] It should be noted that any process or method description in the flowchart or otherwise described herein can be understood as representing a module, segment, or portion of code comprising one or more executable instructions for implementing a particular logical function or process, and the scope of the preferred embodiments of this application includes additional implementations in which functions may be performed not in the order shown or discussed, including in substantially the same manner or in the reverse order of the functions involved, as should be understood by those skilled in the art to which the embodiments of this application pertain.

[0103] It should be understood that various parts of this application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented using software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0104] Those skilled in the art will understand that all or part of the steps of the methods in the above embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, and when executed, the program includes one or a combination of the steps of the method embodiments.

[0105] Furthermore, the functional units in the various embodiments of this application can be integrated into a processing module, or each unit can exist physically separately, or two or more units can be integrated into a module. The integrated module can be implemented in hardware or as a software functional module. If the integrated module is implemented as a software functional module and sold or used as an independent product, it can also be stored in a computer-readable storage medium.

[0106] The storage media mentioned above can be read-only memory, disk, or optical disk, etc.

[0107] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of this application.

Claims

1. A flexible perovskite robot based on environmental perception and autonomous phototactic charging, characterized in that, The robot includes: A multimodal environment perception module is used to monitor ambient lighting information, three-dimensional environmental information, and robot posture status information. An ambient lighting map construction module generates a lighting map based on information collected by a multimodal environment perception module. An autonomous charging decision module outputs the optimal charging strategy based on a lighting map. A dynamic attitude optimization module, which performs charging according to the optimal charging strategy and dynamically adjusts its own attitude in real time during the charging process; A multimodal energy management module is connected to a flexible perovskite power generation module, a lithium polymer battery pack, and a supercapacitor module. The flexible perovskite power generation module covers the surface of the robot's outer shell. The main controller is communicatively connected to the multimodal environment perception module, the ambient lighting map construction module, the autonomous charging decision module, the dynamic attitude optimization module, and the multimodal energy management module.

2. The flexible perovskite robot based on environmental perception and autonomous phototactic charging according to claim 1, characterized in that, The multimodal environment perception module includes a distributed illumination sensor array, a 3D vision unit, and an inertial measurement unit.

3. The flexible perovskite robot based on environmental perception and autonomous phototactic charging according to claim 1, characterized in that, The flexible perovskite power generation module includes several flexible perovskite battery units, each of which is connected to an MPPT controller, and each of the flexible perovskite battery units is distributed in different parts of the robot's body.

4. The flexible perovskite robot based on environmental perception and autonomous phototactic charging according to claim 1, characterized in that, The robot also includes a robot payload, a robot walking mechanism, and a robot servo mechanism. The robot payload, robot walking mechanism, and robot servo mechanism are electrically connected to the multimodal energy management module, and the robot walking mechanism is communicatively connected to the autonomous charging decision module. The robot servo mechanism is communicatively connected to the dynamic posture optimization module.

5. A control method for a flexible perovskite robot based on environmental perception and autonomous phototactic charging, characterized in that, The control method includes: S1: Collect environmental perception data and robot status data; S2: Based on environmental perception data, a semantic map of illumination is generated using SLAM technology and Gaussian process regression algorithm; S3: When the battery level is below the threshold, the task is interrupted and the optimal charging strategy is output through a multi-objective optimization algorithm based on the illumination semantic map and robot state data. S4: The robot navigates to the target charging location according to the optimal charging strategy; S5: Initiate dynamic attitude optimization and perform autonomous charging. When the battery level reaches the threshold, restart the interrupt task.

6. The control method for a flexible perovskite robot based on environmental perception and autonomous phototactic charging according to claim 5, characterized in that, The control method also includes: S6: Dynamically monitors charging power during the charging process and automatically switches between different power supply modes based on the charging power through the energy management unit.

7. The control method for a flexible perovskite robot based on environmental perception and autonomous phototactic charging according to claim 5, characterized in that, The environmental perception data includes: ambient lighting data, 3D vision data, and robot posture data; The robot status data includes: robot battery status, task priority, robot energy consumption cost, robot charging efficiency, and user behavior patterns.

8. The control method for a flexible perovskite robot based on environmental perception and autonomous phototactic charging according to claim 5, characterized in that, The generation logic of the illumination semantic map is as follows: S201: Constructing an environmental spatial structure map based on SLAM technology; S202: Generating a light intensity distribution map using a Gaussian process regression algorithm; S203: Integrate the light intensity distribution map into the time dimension and predict the light time variation pattern based on the Gaussian process regression learning model; S204: Integrate environmental spatial structure map, light intensity distribution map, and light time variation pattern with semantic information to generate a light semantic map.

9. The control method for a flexible perovskite robot based on environmental perception and autonomous phototactic charging according to claim 5, characterized in that, The output logic of the optimal charging strategy is as follows: S301: Acquire robot state data and lighting semantic map; S302: Output the optimal charging strategy through a multi-objective optimization algorithm, and the optimal charging strategy includes: target charging location, optimal movement path and estimated charging time.

10. The control method for a flexible perovskite robot based on environmental perception and autonomous phototactic charging according to claim 5, characterized in that, The logic for dynamic attitude optimization is as follows: S501: Adjust the robot's orientation according to the direction of the light; S502: Adjust the angle of the robot's head and body; S503: Sample the current power generation and determine whether the power generation has increased. If yes, continue to correct at a fixed frequency in the current adjustment direction and execute S502; if no, return to S502 and adjust in the opposite direction. S504: Determine whether the power generation has reached the optimal threshold. If yes, end the adjustment and maintain the current attitude; otherwise, return to S503.

Images