Navigation system of cleaning robot

Abstract

The invention discloses a cleaning robot navigation system, which comprises a central control unit, a sensor module, a motion control system, a cleaning assembly, a communication assembly, a safety and diagnosis module and a battery and power supply management module, and is characterized in that the sensor module comprises a laser radar, an ultrasonic sensor, a visual camera and an inertial measurement unit; the laser radar is used for environment scanning and obstacle detection, the ultrasonic sensor is used for close-range obstacle detection, the visual camera is used for image processing and ground analysis, and the inertial measurement unit is used for attitude and position detection. A series of innovative technologies such as advanced sensor technology, data fusion, path planning, dynamic obstacle avoidance, cleaning task optimization and the like are combined, and a plurality of problems faced by an existing cleaning robot are solved. According to the technical scheme, the cleaning robot can be remarkably improved in the aspects of environment perception, navigation precision, cleaning efficiency, task optimization and the like.

Description

Technical Field

[0001] The present invention relates to the technical field of cleaning robots, and in particular to a cleaning robot navigation system. Background Art

[0002] With the rapid development of smart homes, automated cleaning equipment is increasingly being used in homes, offices and commercial spaces. Traditional manual cleaning methods are not only time-consuming and laborious, but the cleaning effect is often unsatisfactory, especially in hard-to-reach areas such as corners and under furniture. As an emerging automation product, cleaning robots can efficiently and automatically clean floors through intelligent navigation and path planning, greatly improving cleaning efficiency and quality.

[0003] Most cleaning robots on the market currently have low navigation accuracy: many cleaning robots rely on simple collision sensors or preset paths for cleaning, cannot adapt to complex environments, and are prone to missing or repeating cleaning. Poor obstacle avoidance: robots are prone to collision or getting stuck when facing dynamic environments or complex obstacles, resulting in work interruptions. Single cleaning strategy: most existing cleaning robots rely on preset cleaning modes and lack intelligent adjustment capabilities for different floors and stain types, resulting in unstable cleaning effects. Insufficient battery life: Due to battery capacity limitations, many robots need to be charged multiple times when cleaning large areas, reducing work efficiency. Summary of the invention

[0004] The purpose of the present invention is to provide a cleaning robot navigation system. In order to solve these problems, the intelligent cleaning robot needs to have a more intelligent navigation and control system that can efficiently plan paths, avoid obstacles, respond to environmental changes, and select the best cleaning mode according to different cleaning needs.

[0005] To achieve the above object, the present invention is implemented according to the following technical solutions:

[0006] The present invention includes a central control unit, a sensor module, a motion control system, a cleaning component, a communication component, a safety and diagnostic module, and a battery and power management module. The central control unit is responsible for data processing, decision-making and control; the sensor module collects the robot status and environment, the motion control system is responsible for controlling the movement of the robot, the cleaning component includes a vacuum system, a wet mop system and a roller brush to perform actual cleaning tasks, and the communication component realizes remote control and data transmission; the safety and diagnostic module monitors the health status of the robot, performs collision detection and fault diagnosis, and the battery and power management module includes a battery, a charging interface and a battery monitoring module to ensure that the system has sufficient power to operate and automatically charge.

[0007] The sensor module includes a laser radar, an ultrasonic sensor, a visual camera, and an inertial measurement unit. The laser radar performs environment scanning and obstacle detection, the ultrasonic sensor is used for close-range obstacle detection, the visual camera is used for image processing and ground analysis, and the inertial measurement unit is used for posture and position detection.

[0008] The central control unit integrates sensor data processing algorithms, path planning algorithms, obstacle avoidance algorithms, positioning and map building, and cleaning task optimization algorithms.

[0009] The sensor data processing algorithm combines the observations from multiple sensors to estimate the position and velocity of the robot, which is achieved through the Kalman filter algorithm and the particle filter algorithm; the Kalman filter algorithm predicts the next position of the robot based on the current state estimate and motion model; uses the actual measurement value to update the predicted position, and adjusts the estimated value through the Kalman gain; and obtains the optimal position and posture estimate of the robot by fusing multiple sensor data;

[0010] The particle filter algorithm initializes multiple particles by giving an initial state estimate; predicts each particle based on the robot motion model; updates the particle weights using sensor data; selects particles with high weights through a resampling strategy, eliminates particles with smaller weights, and updates the state estimate.

[0011] The path planning algorithm includes global path planning and local path planning;

[0012] The global path planning is as follows: starting from the starting point, initialize the distance of all nodes to infinity and the distance of the starting point to 0; each time select the unvisited node closest to the starting point and update the shortest path of its adjacent nodes; repeat this process until all nodes are visited;

[0013] Local path planning is as follows: calculate the possible velocity space of the robot at the current moment, generate possible motion trajectories based on the robot's dynamic constraints, use the cost function to evaluate the feasibility of each trajectory, and select the optimal trajectory for execution.

[0014] The cleaning task optimization algorithm is as follows: the robot obtains status and reward signals by interacting with the environment; the robot updates its Q value table and is trained through a neural network to improve future decisions; after multiple trainings, the robot learns to select the optimal cleaning strategy in different scenarios.

[0015] The obstacle avoidance algorithm predicts the location and path of obstacles by training a deep neural network and optimizes the obstacle avoidance strategy by combining reinforcement learning.

[0016] The beneficial effects of the present invention are:

[0017] The present invention is a cleaning robot navigation system. Compared with the prior art, the present invention combines a series of innovative technologies such as advanced sensor technology, data fusion, path planning, dynamic obstacle avoidance, and cleaning task optimization, aiming to solve multiple problems faced by existing cleaning robots. Through this technical solution, the cleaning robot can achieve significant improvements in environmental perception, navigation accuracy, cleaning efficiency, and task optimization. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 It is a system structure principle block diagram of the present invention. DETAILED DESCRIPTION

[0019] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments. The illustrative embodiments and descriptions of the present invention are used to explain the present invention but are not intended to limit the present invention.

[0020] like Figure 1 As shown: the present invention includes a central control unit, a sensor module, a motion control system, a cleaning component, a communication component, a safety and diagnostic module, and a battery and power management module. The central control unit is responsible for data processing, decision-making and control; the sensor module collects the robot status and environment, the motion control system is responsible for controlling the movement of the robot, the cleaning component includes a vacuum system, a wet mop system and a roller brush to perform actual cleaning tasks, and the communication component realizes remote control and data transmission; the safety and diagnostic module monitors the health status of the robot, performs collision detection and fault diagnosis, and the battery and power management module includes a battery, a charging interface and a battery monitoring module to ensure that the system has sufficient power to operate and automatically charge.

[0021] The sensor module includes a laser radar, an ultrasonic sensor, a visual camera, and an inertial measurement unit. The laser radar performs environment scanning and obstacle detection, the ultrasonic sensor is used for close-range obstacle detection, the visual camera is used for image processing and ground analysis, and the inertial measurement unit is used for posture and position detection.

[0022] The central control unit is the core of the system and is responsible for data processing, decision-making and control. It interacts with all other modules for data. The sensor module includes multiple sensors, LiDAR: for environment scanning and obstacle detection. Ultrasonic sensor: for close-range obstacle detection. Visual sensor (camera): for image processing and ground analysis. IMU (inertial measurement unit): for attitude and position detection.

[0023] Motion control system: Contains electric motors and drive wheels, responsible for controlling the movement of the robot.

[0024] Cleaning components: including the vacuum system, wet mop system and roller brush, perform the actual cleaning tasks.

[0025] Communication component: Connect to the cloud platform or mobile application via Wi-Fi to achieve remote control and data transmission.

[0026] Battery and power management module: includes battery, charging interface and battery monitoring module to ensure that the system has enough power to operate and automatically charge.

[0027] Safety and diagnostic module: monitors the robot's health status, performs collision detection and fault diagnosis.

[0028] The central control unit integrates sensor data processing algorithms, path planning algorithms, obstacle avoidance algorithms, positioning and map construction, and cleaning task optimization algorithms. Cleaning robots usually use multiple sensors (such as lidar, cameras, ultrasonic sensors, IMUs, etc.) for environmental perception and self-positioning. In order to improve positioning accuracy and path planning reliability, the data from these sensors needs to be fused.

[0029] The sensor data processing algorithm combines the observations from multiple sensors to estimate the position and velocity of the robot, which is achieved through the Kalman filter algorithm and the particle filter algorithm; the Kalman filter algorithm predicts the next position of the robot based on the current state estimate and motion model; uses the actual measurement value to update the predicted position, and adjusts the estimate value through the Kalman gain; obtains the optimal position and posture estimate of the robot by fusing multiple sensor data; and is suitable for processing the state estimation of linear or approximately linear systems. In SLAM (Simultaneous Localization and Mapping), the Kalman filter algorithm locates the robot and processes sensor data such as IMU, GPS, and LiDAR.

[0030] The particle filter algorithm initializes multiple particles by giving an initial state estimate; predicts each particle based on the robot motion model; updates the particle weights using sensor data; selects particles with high weights through a resampling strategy, eliminates particles with small weights, and updates the state estimate. Particle filtering is a nonlinear, non-Gaussian filtering algorithm based on the Monte Carlo method, which is suitable for complex dynamic environments. Compared with EKF, PF can handle more complex environments and can adapt to data from multiple sensors.

[0031] In the cleaning robot system, algorithms involving data processing, decision-making and control are very critical. Different algorithms are responsible for different tasks, such as sensor data processing, path planning, obstacle avoidance, positioning and map construction, cleaning task optimization, etc.

[0032] The path planning algorithm includes global path planning and local path planning;

[0033] Global path planning is as follows: starting from the starting point, initialize the distance of all nodes to infinity and the distance of the starting point to 0; each time select the unvisited node closest to the starting point and update the shortest path of its adjacent nodes; repeat this process until all nodes are visited; the purpose of global path planning is to find the shortest path or optimal path from the starting point to the end point on a known map. It is used to find the shortest path from the source node to the target node in the graph. In the absence of a heuristic function, global path planning compares all possible paths and selects the path with the lowest cost.

[0034] Local path planning is: calculate the possible velocity space of the robot at the current moment, generate possible motion trajectories based on the robot's dynamic constraints, use the cost function to evaluate the feasibility of each trajectory, and select the optimal trajectory for execution. Local path planning refers to how the robot avoids obstacles in real time during movement. The optimal control input is selected taking into account the robot's dynamic constraints such as speed and acceleration. Local path planning is used for dynamic obstacle avoidance and motion control.

[0035] The cleaning task optimization algorithm is as follows: the robot obtains status and reward signals by interacting with the environment; the robot updates its Q value table and is trained through a neural network to improve future decisions; after multiple trainings, the robot learns to choose the optimal cleaning strategy in different scenarios. In cleaning tasks, in addition to path planning, intelligent decisions are also required on the cleaning area and cleaning mode (vacuuming, mopping, etc.). The cleaning task optimization algorithm is able to learn the optimal strategy in a large-scale state space. In cleaning tasks, robots can use the cleaning task optimization algorithm to learn how to choose the appropriate cleaning method (such as whether to use a wet mop, roller brush, etc.) according to the ground state.

[0036] The obstacle avoidance algorithm predicts the location and path of obstacles by training a deep neural network and optimizes the obstacle avoidance strategy by combining reinforcement learning.

[0037] High-precision positioning and navigation:

[0038] The multi-sensor fusion of LiDAR, ultrasonic sensors, visual sensors and IMU (Inertial Measurement Unit) can efficiently build high-precision maps and achieve accurate positioning and path planning. The robot can navigate seamlessly in complex environments to avoid missed and repeated cleaning.

[0039] Through data fusion algorithms such as **Extended Kalman Filter (EKF) or Particle Filter (PF)**, the positioning accuracy and robustness are further improved, especially in the case of incomplete or partial loss of multi-sensor data, ensuring the stable operation of the robot.

[0040] Intelligent path planning and obstacle avoidance capabilities:

[0041] The A* algorithm and Dijkstra algorithm are used for global path planning, which can calculate the shortest or optimal path based on a known map to ensure that the robot can clean the entire area efficiently.

[0042] In a dynamic environment, the dynamic window method (DWA) and artificial potential field method (APF) are used for local path planning to adjust the cleaning path in real time to avoid new obstacles or dynamic obstacles.

[0043] The robot can adjust its motion trajectory in real time based on sensor feedback to avoid collisions or jams, greatly improving the safety and efficiency of movement.

[0044] Flexible cleaning task management and optimization:

[0045] Reinforcement learning (RL) algorithms and heuristic algorithms (such as simulated annealing, genetic algorithm, etc.) are used to optimize cleaning strategies. The robot can adaptively adjust the cleaning mode (such as wet mopping, vacuuming, roller brushing, etc.) according to different cleaning areas, floor types and stain distribution to ensure the maximum cleaning effect.

[0046] The reinforcement learning model can optimize itself based on historical experience, gradually improve its adaptability and processing efficiency to various environments, and reduce human intervention.

[0047] Intelligent battery management and battery life optimization:

[0048] The introduction of the battery management module monitors the battery status in real time to ensure that the robot can automatically return to the charging station for charging when the battery is low. When charging is completed, the robot can continue to perform the unfinished cleaning task, maximizing working time and cleaning efficiency.

[0049] The battery life has been effectively improved, supporting large-area cleaning and reducing the time consumed by multiple charging.

[0050] Multi-task collaboration and remote control:

[0051] Through the combination of multi-robot communication and cloud platform, multiple robots can work together to complete the cleaning tasks of large areas and reduce cleaning time.

[0052] Users can remotely control the robot through mobile applications or cloud platforms, check the cleaning status, set cleaning tasks or optimize cleaning paths, making the cleaning process more flexible and convenient.

[0053] Improved user experience:

[0054] It provides multiple cleaning modes (such as deep cleaning, quick cleaning, fixed-point cleaning, etc.) and personalized settings to meet the cleaning needs of different users.

[0055] The robot can automatically identify the floor material (such as wooden floors, carpets, tiles, etc.) and automatically adjust the cleaning mode according to the material to ensure the best cleaning effect.

[0056] Fault diagnosis and early warning mechanism: the robot can automatically detect and alarm abnormal conditions such as sensor failure, low battery, etc., to avoid cleaning interruptions due to faults.

[0057] To sum up, this technical solution not only solves the shortcomings of existing cleaning robots in path planning, obstacle avoidance, cleaning effect, endurance, etc., but also improves the robot's adaptability and cleaning efficiency by introducing intelligent algorithms (such as reinforcement learning, sensor data fusion, etc.), meeting the cleaning needs in different scenarios, and has high market application value and technological innovation.

[0058] Example: Cleaning robot navigation system based on multi-sensor fusion and intelligent path planning

[0059] The following is an implementation example of a cleaning robot navigation system based on the present technical solution, which describes in detail how the robot uses sensor data fusion, path planning, obstacle avoidance, cleaning task management and other technologies in different environments to achieve efficient cleaning operations.

[0060] 1. System composition

[0061] The cleaning robot in this embodiment includes the following main modules:

[0062] Central control unit: responsible for sensor data processing, path planning, decision control and task management.

[0063] Sensor module: includes LiDAR, ultrasonic sensor, visual camera, IMU (inertial measurement unit) and other sensors for environmental perception and positioning.

[0064] Battery and power management module: includes battery, charging interface and charging station, responsible for power supply and management.

[0065] Motion control system: includes electric motors, drive wheels, etc., used to control the movement of the robot.

[0066] Cleaning module: includes vacuum system, wet mop system and roller brush, etc., to perform floor cleaning tasks.

[0067] Communication module: realize wireless communication with user mobile phones, cloud servers, etc., and support remote control and status synchronization.

[0068] 2. Workflow

[0069] 2.1 Startup and self-test

[0070] When the user starts the cleaning robot, the robot first performs a system self-check to check the working status of core components such as batteries, motors, sensors, etc. After passing the self-check, the robot enters standby mode, waiting for the user to set a cleaning task or automatically start according to a predetermined plan.

[0071] 2.2 Environmental Perception and Map Construction

[0072] Sensor data collection:

[0073] The laser radar (LiDAR) performs 360-degree scanning to obtain distance data of the surrounding environment; the ultrasonic sensor is used for close-range obstacle detection; the visual camera cooperates with the deep learning algorithm to detect ground stains and ground types (such as floors and carpets); the IMU sensor provides position information and posture data.

[0074] Sensor data fusion: Through the **Extended Kalman Filter (EKF)** algorithm, the robot can estimate its current position in real time and build an environmental map by fusing data from LiDAR, IMU, ultrasonic and visual cameras. In this way, the robot can know its position in the cleaning area and form a detailed environmental map, including the location of obstacles, furniture and other objects.

[0075] 2.3 Path planning and cleaning task allocation

[0076] Global path planning: Use the A* algorithm to perform global path planning on the environment map to determine the shortest path from the starting point to the target area. Path planning takes into account the location of obstacles and selects an effective cleaning route. This process is based on a known environment map to ensure that the robot can cover the entire area.

[0077] Local obstacle avoidance: During the cleaning process, the robot may encounter dynamic or unknown obstacles (such as moving furniture or pets). At this time, the robot uses the **Dynamic Window Algorithm (DWA)** for local path planning, and avoids sudden obstacles by scanning the surrounding environment in real time, and continues to perform the cleaning task.

[0078] Cleaning task management and optimization: The robot automatically analyzes the cleaning needs of each area based on the environmental map. The robot will identify dirty areas and clean them first, or automatically adjust the cleaning strategy according to the preset cleaning mode (such as deep cleaning, fixed-point cleaning, etc.). Based on reinforcement learning (RL), the robot can gradually learn the optimal cleaning strategy in different tasks.

[0079] 2.4 Cleaning process control and execution

[0080] Cleaning mode selection: The robot automatically selects the appropriate cleaning mode based on the floor material (such as wooden floor, tile or carpet). For example, the robot uses the vacuum function on wooden floor, and activates the roller brush function and wet mops on carpet. During the cleaning process, the robot detects the dirtiness of the floor in real time and uses visual sensors to determine whether the cleaning intensity needs to be increased.

[0081] Cleaning module operation:

[0082] Dust suction system: During the cleaning process, the dust suction motor uses strong suction to absorb ground garbage and store it in the dust box.

[0083] Wet mop system: When the robot enters an area that requires wet mopping, the wet mop system is automatically activated, water is distributed to the wet cloth through a water pump, and the mopping operation is performed. The amount of water in the wet mop system is dynamically adjusted according to the type of floor and the degree of dirtiness.

[0084] Roller brush system: When cleaning the junction of carpet and floor, the robot will activate the roller brush based on sensor feedback to clean hair and other stubborn stains on the ground.

[0085] 2.5 Battery management and battery life optimization

[0086] Power monitoring and charging management: The robot monitors the battery power in real time during the cleaning process. If the power is lower than the preset threshold, the robot will automatically return to the charging base station to charge, and resume the cleaning task after charging is completed, avoiding downtime or missed cleaning.

[0087] Battery life optimization: Based on the battery status and workload, the robot will intelligently adjust the movement speed and cleaning mode to optimize the battery life. For example, when wet mopping is not needed, the robot may only use the vacuuming function to extend the cleaning time.

[0088] 2.6 Handling Failures or Exceptions

[0089] During the operation, the robot uses the safety and diagnostic module to monitor the health of sensors, motors, batteries and other systems in real time. Once an abnormality is detected (such as sensor failure, obstacle blocking, battery exhaustion, etc.), the robot will sound an alarm and perform fault handling. Fault handling includes:

[0090] Stop cleaning and return to the base station.

[0091] Send fault information to the user's mobile app.

[0092] Automatically adjust or repair the system based on the diagnosis results.

[0093] 2.7 Multi-task collaboration and remote control

[0094] Multi-robot collaboration: In large-area cleaning places (such as shopping malls, offices, etc.), multiple cleaning robots can work together and share location information, obstacle information and task allocation. Through the cloud platform or local area network, multiple robots can coordinate operations to reduce cleaning time and duplication of work.

[0095] Remote control and monitoring: Users can remotely start, pause or adjust cleaning tasks and view real-time maps and cleaning progress through mobile phone applications or cloud platforms. Users can also set scheduled cleaning and select cleaning areas according to specific needs, and the robot will perform corresponding tasks according to these instructions.

[0096] 3. Effects of the Embodiment

[0097] Through this system, cleaning robots can efficiently perform cleaning tasks in different environments, solving the problems encountered by existing robot cleaning, such as positioning error, inaccurate path planning, untimely obstacle avoidance, unstable cleaning effect, etc. The specific effects are as follows:

[0098] Efficient cleaning coverage: The robot can ensure efficient coverage of the entire cleaning area based on accurate environmental maps and path planning algorithms, avoiding missed or repeated cleaning.

[0099] Dynamic obstacle avoidance capability: Through local path planning algorithms (such as dynamic window method and artificial potential field method), the robot can avoid obstacles in real time to ensure that the cleaning task is not interrupted.

[0100] Intelligent cleaning mode: Based on reinforcement learning and sensor feedback, the robot can automatically adjust the cleaning strategy according to different floor types and dirtiness, providing more efficient and energy-saving cleaning services.

[0101] Long battery life and self-charging function: The robot can intelligently manage battery life and automatically return to the charging base station to recharge when the battery is low, ensuring uninterrupted work.

[0102] The cleaning robot of this embodiment achieves efficient navigation and cleaning in complex environments by comprehensively using technologies such as sensor data fusion, intelligent path planning, dynamic obstacle avoidance, and reinforcement learning, and can automatically optimize cleaning modes and battery life, greatly improving cleaning efficiency and user experience. These technologies make the cleaning robot have broad application prospects in home, commercial, and office environments.

[0103] The technical solution of the present invention is not limited to the above-mentioned specific embodiments. All technical variations made according to the technical solution of the present invention fall within the protection scope of the present invention.

Claims

1. A cleaning robot navigation system, characterized in that: It includes a central control unit, a sensor module, a motion control system, a cleaning component, a communication component, a safety and diagnostic module, and a battery and power management module. The central control unit is responsible for data processing, decision-making and control; the sensor module collects the robot status and environment, the motion control system is responsible for controlling the movement of the robot, the cleaning component includes a vacuum system, a wet mop system and a roller brush to perform actual cleaning tasks, and the communication component realizes remote control and data transmission; the safety and diagnostic module monitors the health status of the robot, performs collision detection and fault diagnosis, and the battery and power management module includes a battery, a charging interface and a battery monitoring module to ensure that the system has sufficient power to operate and automatically charge.

2. The cleaning robot navigation system according to claim 1, characterized in that: The sensor module includes a laser radar, an ultrasonic sensor, a visual camera, and an inertial measurement unit. The laser radar performs environment scanning and obstacle detection, the ultrasonic sensor is used for close-range obstacle detection, the visual camera is used for image processing and ground analysis, and the inertial measurement unit is used for posture and position detection.

3. The cleaning robot navigation system according to claim 2, characterized in that: The central control unit integrates sensor data processing algorithms, path planning algorithms, obstacle avoidance algorithms, positioning and map building, and cleaning task optimization algorithms.

4. The cleaning robot navigation system according to claim 3, characterized in that: The sensor data processing algorithm combines the observations from multiple sensors to estimate the position and velocity of the robot, which is achieved through the Kalman filter algorithm and the particle filter algorithm; the Kalman filter algorithm predicts the next position of the robot based on the current state estimate and motion model; uses the actual measurement value to update the predicted position, and adjusts the estimated value through the Kalman gain; and obtains the optimal position and posture estimate of the robot by fusing multiple sensor data; The particle filter algorithm initializes multiple particles by giving an initial state estimate; predicts each particle based on the robot motion model; updates the particle weights using sensor data; selects particles with high weights through a resampling strategy, eliminates particles with smaller weights, and updates the state estimate.

5. The cleaning robot navigation system according to claim 3, characterized in that: The path planning algorithm includes global path planning and local path planning; The global path planning is as follows: starting from the starting point, initialize the distance of all nodes to infinity and the distance of the starting point to 0; each time select the unvisited node closest to the starting point and update the shortest path of its adjacent nodes; repeat this process until all nodes are visited; Local path planning is as follows: calculate the possible velocity space of the robot at the current moment, generate possible motion trajectories based on the robot's dynamic constraints, use the cost function to evaluate the feasibility of each trajectory, and select the optimal trajectory for execution.

6. The cleaning robot navigation system according to claim 3, characterized in that: The cleaning task optimization algorithm is as follows: the robot obtains status and reward signals by interacting with the environment; the robot updates its Q value table and is trained through a neural network to improve future decisions; after multiple trainings, the robot learns to select the optimal cleaning strategy in different scenarios.

7. The cleaning robot navigation system according to claim 3, characterized in that: The obstacle avoidance algorithm predicts the location and path of obstacles by training a deep neural network and optimizes the obstacle avoidance strategy by combining reinforcement learning.

Images