A campus trash-collecting robot

Abstract

A campus waste collection robot, integrating motion prediction technology and real-time obstacle avoidance strategies, and equipped with a computer program, can improve the robot's autonomous navigation capabilities and cleaning efficiency in complex campus environments. The method includes: collecting environmental data using LiDAR and RGB-D cameras; generating a static grid map using SLAM technology and identifying the motion state of dynamic obstacles in real time; predicting the trajectory of dynamic obstacles using a temporal convolutional network and a bidirectional GRU model to generate reachable areas; constructing static, dynamic, and gravitational fields based on an improved artificial potential field method and dynamically adjusting the obstacle avoidance strategy; implementing global path planning using the A* algorithm, optimizing local paths using a dynamic window method, and updating in real time; achieving high-precision path tracking through stepper motor drive combined with a PID control algorithm, triggering an obstacle avoidance mode when encountering emergency obstacles. It is suitable for waste collection operations in complex environments such as campuses, shopping malls, and factories.

Description

Technical Field

[0001] It relates to the field of robotics technology, specifically a campus trash-collecting robot. Background Technology

[0002] With the rapid development of artificial intelligence, robotics, and automation, mobile robots have become indispensable tools in many scenarios. For example, in industrial manufacturing, mobile robots are often used for tasks such as material handling and precision assembly; in the medical field, they are used to transport medical equipment and medicines; and in agriculture, agricultural robots are used for spraying pesticides or sowing seeds in the fields. These applications demonstrate the significant value of mobile robots in improving work efficiency and saving labor costs.

[0003] In campus environments, garbage collection robots are gradually gaining attention as auxiliary tools for environmental cleaning and management. Existing technologies already include some research and applications of garbage collection robots. For example:

[0004] Fixed-path garbage cleaning robots: These robots clean up garbage on the road according to a preset fixed cleaning path. Although they can operate efficiently in enclosed environments (such as industrial plants), they cannot dynamically adapt to environmental changes. In open campus environments, they are prone to stopping or even colliding when encountering sudden obstacles (such as pedestrians or vehicles).

[0005] Cleaning robots based on real-time obstacle avoidance: These robots are equipped with sensors (such as ultrasonic or lidar) that can detect obstacles in real time and avoid them. However, due to the lack of global path planning capabilities, their cleaning process often suffers from local optimization problems, which can easily lead to detours, increased energy consumption, and low task efficiency.

[0006] Dynamic environment cleaning robots: Some robots attempt to use dynamic trajectory prediction technology to predict the movement trend of dynamic obstacles in advance, thereby adjusting their travel route. Although this method can theoretically improve cleaning efficiency, its application in real-world scenarios still faces many challenges due to high algorithm complexity, large computational resource requirements, and imperfect modeling of dynamic obstacles.

[0007] Existing technological research indicates that traditional garbage-collecting robots still face the following major problems in complex environments:

[0008] Lack of path planning capabilities that combine global and local aspects: Fixed path planning cannot cope with the complexity of dynamic environments, while a single real-time obstacle avoidance strategy can easily lead to unreasonable local paths and low overall cleaning efficiency.

[0009] Insufficient handling of dynamic obstacles: Existing robots have limited ability to perceive dynamic obstacles and cannot predict the movement trajectory of obstacles in advance. Obstacle avoidance behavior is often delayed, which may cause safety problems.

[0010] Insufficient energy consumption optimization: Due to the lack of path optimization algorithms, existing robots are prone to redundant movements, leading to unnecessary increases in energy consumption. Summary of the Invention

[0011] To address the technical problems of existing dynamic environment cleaning robots, such as the lack of global and local path planning capabilities and insufficient dynamic obstacle handling, the technical solution provided by this invention includes:

[0012] A method for planning the cleaning route of a campus trash-collecting robot includes:

[0013] The steps include constructing a time-series input of dynamic obstacles based on sensor data and generating the predicted reachable area of ​​the dynamic obstacles;

[0014] The step of constructing a composite potential field including a static potential field, a dynamic potential field, and a gravitational field based on the predicted reachable area of ​​the dynamic obstacle;

[0015] The steps include generating a global path based on the synthesized potential field and dynamically updating the global path based on real-time sensor data.

[0016] Furthermore, a preferred embodiment is provided in which the future trajectory of the dynamic obstacle is predicted in the predicted reachable region of the dynamic obstacle using a temporal convolutional network and a bidirectional GRU model.

[0017] Furthermore, a preferred embodiment is provided in which the static potential field calculates the repulsive force value based on the geometric characteristics of the obstacle.

[0018] Furthermore, a preferred embodiment is provided in which the dynamic potential field generates a repulsion region using the predicted trajectory of dynamic obstacles and dynamically adjusts the weights.

[0019] Furthermore, a preferred embodiment is provided in which the gravitational field is controlled by a Lyapunov function to generate a smooth gravitational gradient.

[0020] Furthermore, a preferred implementation method is provided, which generates a global path through the A* algorithm and optimizes the local path by combining the dynamic window method, and dynamically updates the path based on real-time sensor data.

[0021] Based on the same inventive concept, this invention also provides a campus waste collection robot, including a mobile platform and a robotic arm. The mobile platform carries the robotic arm and enables the robot to move freely in the campus environment. The robotic arm picks up waste and places it onto a waste collection device on the mobile platform. The robot is used to implement the method described, including:

[0022] The mobile platform includes a rear-wheel drive system and a front-wheel steering system. The rear-wheel drive system is driven by a stepper motor to rotate the rear wheels through chain transmission. The front-wheel steering system is based on the Ackermann steering geometry principle and is connected to the stepper motor through a coupling to achieve precise steering.

[0023] The robotic arm is mounted on the top of the mobile platform and includes an upper arm, a forearm, and a gripper. The upper arm and forearm rotate through gear transmission, and the gripper is driven to lift and lower through a screw mechanism.

[0024] Based on the same inventive concept, the present invention also provides a computer storage medium for storing a computer program, wherein when the computer program is read by a computer, the computer executes the method described thereon.

[0025] Based on the same inventive concept, the present invention also provides a computer, including a processor and a storage medium, wherein when the processor reads a computer program stored in the storage medium, the computer executes the method described thereon.

[0026] Based on the same inventive concept, the present invention also provides a computer program product, which, when executed, implements the method described.

[0027] Compared with the prior art, the advantages of the technical solution provided by the present invention are as follows:

[0028] By combining motion prediction technology and real-time obstacle avoidance strategies, the path planning capability of the garbage-collecting robot in dynamic environments has been significantly improved. Compared with traditional robots that rely solely on real-time obstacle avoidance, this solution uses a temporal convolutional network (TCN) and a bidirectional GRU model to predict the future trajectory of dynamic obstacles, making obstacle avoidance behavior more proactive and intelligent, reducing lag issues in path adjustment, and greatly improving the robot's cleaning efficiency and safety in dynamic campus environments.

[0029] Based on an improved artificial potential field method, a dynamic potential field is generated in the predicted area of ​​dynamic obstacles and then weighted and synthesized with the static potential field and the gravitational field, achieving an organic combination of global path planning and local path optimization. Compared with traditional fixed path planning methods, this approach can dynamically adjust the cleaning path, avoiding stagnation or detours caused by environmental changes, and ensuring the continuity and stability of the cleaning task.

[0030] This paper employs the A* algorithm for global path planning, combined with the Dynamic Window Method (DWA) for local path optimization, enabling the robot to balance global objectives with local environmental adaptability. Compared to robots that rely solely on global path planning, this approach better adapts to unexpected obstacles in complex environments, effectively reduces redundant movements, improves the rationality of path planning, and lowers energy consumption.

[0031] By integrating sensing modules with environmental modeling technology, high-precision static and dynamic obstacle perception is achieved. Compared to solutions relying on a single sensor, this technology significantly improves the comprehensiveness and accuracy of environmental information acquisition through multimodal data fusion from RGB-D cameras, LiDAR, and ultrasonic sensors, providing reliable foundational data for subsequent path planning and obstacle avoidance strategies.

[0032] By using stepper motors combined with PID control algorithms, the accuracy and stability of path tracking during robot execution are ensured. Compared with traditional simple speed control methods, this approach significantly reduces yaw or repetitive movements caused by error accumulation during cleaning, further improving work efficiency.

[0033] Suitable for garbage cleaning and collection in complex environments. Attached Figure Description

[0034] Figure 1 This is a flowchart of the path planning method.

[0035] Figure 2 This is a schematic diagram of the mechanism of a campus trash-collecting robot.

[0036] Figure 3 This is a 3D schematic diagram of a robotic arm. Detailed Implementation

[0037] To make the advantages and benefits of the technical solution provided by the present invention clearer, the technical solution provided by the present invention will now be described in further detail with reference to the accompanying drawings, specifically:

[0038] Implementation Method 1: This implementation method provides a method for planning the cleaning route of a campus garbage collection robot, including:

[0039] The steps include constructing a time-series input of dynamic obstacles based on sensor data and generating the predicted reachable area of ​​the dynamic obstacles;

[0040] The step of constructing a composite potential field including a static potential field, a dynamic potential field, and a gravitational field based on the predicted reachable area of ​​the dynamic obstacle;

[0041] The steps include generating a global path based on the synthesized potential field and dynamically updating the global path based on real-time sensor data.

[0042] Within the predicted reachable region of the dynamic obstacle, the future trajectory of the dynamic obstacle is predicted using a temporal convolutional network and a bidirectional GRU model.

[0043] The static potential field calculates the repulsive force value based on the geometric characteristics of the obstacle.

[0044] The dynamic potential field generates a repulsion region using the predicted trajectory of dynamic obstacles and dynamically adjusts the weights.

[0045] The gravitational field is controlled by a Lyapunov function to generate a smooth gravitational gradient.

[0046] The global path is generated using the A* algorithm, and the local path is optimized using the dynamic window method. The path is then dynamically updated based on real-time sensor data.

[0047] like Figure 2 and 3 As shown, a campus trash-collecting robot is also provided, including a mobile platform and a robotic arm. The mobile platform carries the robotic arm and enables the robot to move freely in the campus environment. The robotic arm picks up trash and places it onto a trash collection device on the mobile platform. The robot is used to implement the method described above, including:

[0048] The mobile platform includes a rear-wheel drive system and a front-wheel steering system. The rear-wheel drive system is driven by a stepper motor to rotate the rear wheels through chain transmission. The front-wheel steering system is based on the Ackermann steering geometry principle and is connected to the stepper motor through a coupling to achieve precise steering.

[0049] The robotic arm is mounted on the top of the mobile platform and includes an upper arm, a forearm, and a gripper. The upper arm and forearm rotate through gear transmission, and the gripper is driven to lift and lower through a screw mechanism.

[0050] Specifically, including:

[0051] 1. Overall Appearance

[0052] The robot consists of a mobile platform and a robotic arm, and is primarily designed for picking up litter on campus. The mobile platform carries the robotic arm, which moves freely via rear-wheel drive and front-wheel steering. The robotic arm has flexible grasping and dropping capabilities to pick up target litter.

[0053] 2. Mobile Platform

[0054] Wheel structure: The mobile platform adopts a four-wheel design, with the rear wheels being drive wheels driven by stepper motors through chain transmission; the front wheels are steering wheels, adopting an Ackermann steering geometry design to ensure flexible steering capability.

[0055] Vehicle frame: The main frame of the platform uses cast aluminum, which ensures structural strength while reducing weight. The vehicle body has a compact design, making it suitable for complex campus environments.

[0056] Transmission System: The rear-wheel drive system transmits the motor's rotation to the rear axle via a sprocket mechanism, which in turn drives the wheels. The sprocket structure design effectively distributes the load on the axle. A reasonable configuration of the center distance and number of teeth ensures smooth transmission.

[0057] 3. Robotic Arm Structure

[0058] Overall Layout: The robotic arm is mounted on top of the mobile platform. It has a column-like structure and three degrees of freedom, including the upper arm joint, forearm joint, and lifting / lowering function of the gripper. The entire robotic arm structure is controlled by stepper motors, which drive gears to complete the movement of each component.

[0059] Boom design: The boom is hinged to the base and can rotate within a 240° range. It is made of aluminum alloy, combining strength and lightness.

[0060] Forearm design: The forearm has a hollow cylindrical structure, and joint rotation is achieved through gear transmission. The hollow design improves bending resistance and reduces structural weight.

[0061] Hand structure: The hand features a gripping design, consisting of two rotating fingers, with a maximum gripping range of 80mm, suitable for picking up common litter (such as bottles and cans). The hand is V-shaped to adapt to the shapes of various objects.

[0062] Drive system: Each joint of the robotic arm is driven by a stepper motor, which, together with a ball screw, enables the lifting and lowering of the gripper, featuring high precision and low error.

[0063] 4. Detailed Features

[0064] Base: The robot's base is made of cast aluminum and is fixedly connected to the mobile platform to provide stable support for the robot.

[0065] Steering mechanism: The steering mechanism of the front wheels is connected to the stepper motor via a coupling, and achieves precise steering based on the Ackermann geometry principle.

[0066] Waste collection function: The mobile platform is equipped with a trash can on the side, and the robotic arm accurately puts the picked-up trash into the trash can.

[0067] Structural optimization: The overall design focuses on simplifying the structure, reducing weight, and improving modularity to facilitate future maintenance and upgrades.

[0068] Specifically, the robotic arm is:

[0069] Overall Structure: This robotic arm adopts a gripping design, using the coordinated action of two grippers to grasp and release objects. Its compact design makes it suitable for installation at the end effector of a robot, enabling the pickup of objects of various shapes.

[0070] Drive mechanism: The robotic arm is driven by a gear transmission system. The main gear is driven by a motor and meshes with the secondary gear to achieve synchronous opening and closing movements of the gripper. The gear transmission ensures the accuracy and stability of the gripper's movements.

[0071] Linkage system: The gripper of the robotic arm is connected to the gear system via multiple links. The linkage mechanism transmits power, enabling the gripper to achieve smooth opening and closing movements, while enhancing the flexibility of the grasping process.

[0072] Hand design: The hand has a two-finger structure with an end design that adapts to grasping objects of various shapes. The spacing between the claws can be adjusted according to the size of the target object, making it suitable for grasping lightweight litter such as bottles, cans, and paper in the campus environment.

[0073] Materials and Installation: This robotic arm is made of lightweight, high-strength materials, facilitating load-bearing capacity and reducing overall weight. The robotic arm connects to the robot arm via a bottom mounting interface, ensuring overall stability and grasping accuracy.

[0074] Implementation Method Two: Combination Figure 1-3 This embodiment describes a method for planning the cleaning route of a campus waste-collecting robot that integrates motion prediction technology and real-time obstacle avoidance strategies. By modeling the static and dynamic environment, it combines global path planning and local path optimization to improve the robot's cleaning efficiency in a dynamic campus environment. The entire implementation process includes five main steps: environmental modeling and data processing, motion prediction, real-time obstacle avoidance strategy, cleaning path planning and real-time updates, and execution control.

[0075] Environmental modeling and data processing (input: static and dynamic obstacles in the environment) involves collecting environmental information through sensing modules to build environmental models of static and dynamic obstacles. First, point cloud data of static obstacles is acquired using LiDAR, and a static raster map of the campus is generated using SLAM technology. Then, RGB-D cameras are used to identify the speed and direction information of dynamic obstacles, recording their position and motion status. Finally, the data collected by the sensors is filtered to remove noise, normalized to a unified coordinate system, and converted into a raster map format for use by subsequent modules.

[0076] Motion prediction (input: position, velocity, and time series of dynamic obstacles) predicts the trajectory of dynamic obstacles, improving the foresight of obstacle avoidance behavior. The time series data of the dynamic obstacles is input into a Temporal Convolutional Network (TCN) to extract its temporal features. Subsequently, a bidirectional GRU model is used to predict the obstacle's trajectory over multiple steps, generating the reachable region of the dynamic obstacle. The priority of the predicted region is dynamically adjusted using an elliptical envelope approach. The prediction results serve as input to the real-time obstacle avoidance strategy.

[0077] The real-time obstacle avoidance strategy (input: environmental model and predicted trajectories of dynamic obstacles) is based on an improved artificial potential field method, constructing a composite potential field of static, dynamic, and gravitational fields. The static potential field calculates the repulsive force value using the geometric features of the obstacle and the Sigmoid function; the dynamic potential field is generated using the predicted region of the dynamic obstacle, with its weight decreasing with each time step; and the gravitational field generates gradient gravity values ​​based on the target point's position. After the three potential fields are weighted and synthesized, the robot's obstacle avoidance strategy is adjusted in real time to ensure the safety and continuity of the cleaning task.

[0078] The cleaning path planning and real-time updates combine global path planning with local path optimization to achieve efficient robot cleaning path planning. First, the A* algorithm is used to generate the globally optimal path on a static map. Then, the local path is optimized using the Dynamic Window Method (DWA), adjusting the obstacle avoidance area based on real-time environmental information. The planning results are dynamically corrected based on sensor updates 10 times per second. If the trajectory deviates from the predicted value, a dynamic potential field is regenerated and the path is updated.

[0079] The robot executes cleaning tasks according to a planned path. The speed and direction of the moving platform are controlled by stepper motors and a servo system, combined with a PID control algorithm to achieve high-precision path tracking. The robotic arm is responsible for picking up trash, utilizing a three-degree-of-freedom joint design to grasp and dispose of it. If an emergency obstacle is detected, an emergency obstacle avoidance mode is triggered, prioritizing bypassing the obstacle before continuing the task.

[0080] Among them: Environmental modeling and data processing: Environmental information modeling is the foundation of the entire system. Through the collaboration of LiDAR and cameras, not only can static maps be generated, but the shape, speed, and direction of dynamic obstacles can also be identified. The grid map generated using SLAM technology is updated in real time, ensuring that the environmental model can reflect changes in the current scene. The data filtering step employs the Kalman filter algorithm to enhance data reliability.

[0081] Motion Prediction: Time series analysis is used to predict the trajectory of dynamic obstacles, ensuring the accuracy of future trajectory points. The Adam optimizer is used to train the TCN and GRU models, with mean squared error (MSE) as the loss function. Predicted elliptical regions prioritize recent trajectories to avoid redundant calculations introduced by long-term information.

[0082] Real-time obstacle avoidance strategy: The gravitational field in the potential field function is designed based on the Lyapunov function to ensure path smoothness. The dynamic potential field uses a custom decreasing weight function, so that the priority of the predicted trajectory gradually decreases over time. When synthesizing the potential field, a conflict resolution mechanism prioritizes safety.

[0083] Cleaning path planning and real-time updates: The path update mechanism introduces a real-time weight adjustment function, enabling the fusion of global and local paths to adapt to dynamic environmental changes. A multi-threaded architecture ensures synchronized operation of path planning and sensor data updates, improving the overall system's real-time performance and response speed.

[0084] Execution Control: The execution module employs a stepper motor combined with a chain drive structure, which features high control precision and stability. The robotic arm's gripping action is achieved through point-to-point control, and the lifting mechanism uses a ball screw design, which improves lifting accuracy while reducing the system's size and weight.

[0085] Implementation Method 3: Combination Figure 1 This embodiment describes the technical solution provided above in further detail through specific examples. Specifically:

[0086] 1. Technical Overview

[0087] By predicting the future trajectories of dynamic obstacles using motion prediction technology and dynamically adjusting the cleaning path in conjunction with real-time obstacle avoidance strategies, a path planning system based on a combination of global and local approaches is formed. During cleaning tasks, the robot can not only autonomously plan an efficient global cleaning path but also avoid dynamic obstacles in real time, ensuring task continuity and safety.

[0088] 2. System Composition

[0089] (1) Sensing module:

[0090] Implementation: Configure a camera (such as an RGB-D camera), LiDAR, and ultrasonic sensors, and connect them to the robot controller using a CAN bus.

[0091] Function: Real-time collection of environmental information, including the location of static obstacles and the speed and direction of dynamic obstacles.

[0092] (2) Motion prediction module:

[0093] Implementation method: Build a Temporal Convolutional Network (TCN) and a bidirectional GRU model based on a deep learning framework (such as TensorFlow or PyTorch).

[0094] Training data: Collect motion trajectory data of pedestrians, vehicles, etc. in the campus environment to construct a time series training set.

[0095] Function: Input the motion state of a dynamic obstacle and output its future reachable area.

[0096] (3) Real-time obstacle avoidance module:

[0097] Implementation method: Based on the improved artificial potential field method, the repulsive force fields of dynamic and static obstacles are constructed, and the total potential field is calculated by combining the gravitational field.

[0098] Algorithm: Implement a custom dynamic adjustment function to detect and update the repulsion weights of obstacles in real time.

[0099] (4) Path planning module:

[0100] Implementation: Use the graph search-based A* algorithm for global path planning, and combine it with DWA (Dynamic Window Method) to achieve local path optimization.

[0101] Function: Generates the globally optimal path and dynamically updates the obstacle avoidance area.

[0102] (5) Execution control module:

[0103] Implementation method: Stepper motors and servo systems are used, and path tracking is achieved through an embedded controller (such as STM32).

[0104] Function: Adjust the robot's speed and direction according to the planned path.

[0105] 3. Specific Implementation Method

[0106] (1) Environmental modeling and data processing

[0107] Environmental modeling steps:

[0108] Use LiDAR to acquire static environmental point cloud data.

[0109] A static raster map of the campus was generated using SLAM (Simultaneous Localization and Mapping) technology.

[0110] It works in conjunction with a camera to identify moving targets and record their position and speed.

[0111] Tools: Use open-source SLAM frameworks (such as Cartographer or RTAB-Map).

[0112] Data processing steps:

[0113] The raw data collected by the sensor is filtered to remove noise.

[0114] A normalization algorithm is used to map sensor data to a unified coordinate system.

[0115] Convert it into an environmental raster map format for use by subsequent modules.

[0116] (2) Motion prediction

[0117] Dynamic obstacle trajectory prediction steps:

[0118] The time series of dynamic obstacles are fed into a temporal convolutional network (TCN).

[0119] After extracting its temporal features, a bidirectional GRU is used to predict future multi-step motion trajectories.

[0120] Implementation: The Adam optimizer is used during model training, and the mean squared error (MSE) is used as the loss function.

[0121] Specifically:

[0122] Dynamic obstacle trajectory prediction

[0123] Time series input

[0124] Let the historical trajectory of the dynamic obstacle be:

[0125]

[0126] in Indicates location, Represents the velocity component. For timestamps.

[0127] TCN feature extraction involves inputting the time series data into a temporal convolutional network (TCN) to extract temporal features.

[0128]

[0129] in Indicates time eigenvectors.

[0130] GRU predicts future trajectory

[0131] The extracted features are input into a bidirectional GRU to predict future multi-step trajectory points:

[0132]

[0133] in Indicates the future number The predicted point of the step.

[0134] Elliptical dynamic region generation: Generates elliptical regions of dynamic obstacles based on predicted points.

[0135]

[0136] in and These are the major and minor axes of the ellipse, respectively, and are dynamically adjusted based on the speed and direction of the obstacle.

[0137] Steps for generating a dynamic region:

[0138] A multi-step elliptical envelope potential field is generated based on the predicted trajectory.

[0139] The weights are dynamically adjusted to give higher priority to recent regions.

[0140] Tools: Real-time ellipse plotting using NumPy and Matplotlib.

[0141] (3) Real-time obstacle avoidance strategy

[0142] Steps for constructing a static potential field:

[0143] The repulsive force of a static obstacle is calculated using the Sigmoid function.

[0144] Different weights are assigned to the geometric features of obstacles.

[0145] Implementation: Implement a dynamically adjusted Sigmoid function using a custom function.

[0146] Steps for constructing a dynamic potential field:

[0147] The repulsive force field of dynamic obstacles is defined using the predicted elliptical region.

[0148] The repulsion weight is decreased over time to reduce the impact of long-term trajectories on current planning.

[0149] Steps for constructing a gravitational field:

[0150] Generate a gravitational field at the target point.

[0151] The gravitational gradient is controlled using the logarithmic Lyapunov function.

[0152] Implementation: Define a dynamic adjustment function for the target point to ensure a smooth path.

[0153] Steps for synthesizing the total potential field:

[0154] Weighted synthesis of dynamic potential field, static potential field and gravitational field.

[0155] Define a conflict resolution mechanism, prioritizing security.

[0156] Specifically:

[0157] Dynamic potential field generation: Constructing a dynamic potential field using the predicted region of dynamic obstacles.

[0158]

[0159] in: It is the center point of the dynamic obstacle; The weight of the obstacle is dynamically adjusted. , It is a prediction of the time step. Control the rate of weight decay. The current time; It is to avoid small values ​​where the denominator is zero.

[0160] Static potential generation

[0161] Construct a repulsive potential field for a static obstacle:

[0162]

[0163] in: Indicates the robot and the first The distance to a static obstacle.

[0164] Gravitational field generation: Generate a gravitational field at the target point.

[0165]

[0166] in It is the gravitational field strength coefficient.

[0167] The total potential field is synthesized by weighted synthesis of the dynamic potential field, static potential field, and gravitational field:

[0168]

[0169] in: and These are the weight parameters for the dynamic potential field and the static potential field, respectively, which are adjusted according to the real-time environment.

[0170] Obstacle avoidance decision

[0171] Calculate the robot's direction of motion based on the total potential field:

[0172]

[0173] in That is the best direction for the robot to travel.

[0174] (4) Cleaning path planning and real-time updates

[0175] Path planning steps:

[0176] Perform a global A* path search on a static map.

[0177] The path is segmented, and local paths are optimized using DWA.

[0178] Implementation: Path planning is implemented using the ROS MoveBase framework.

[0179] Real-time update steps:

[0180] Obstacle status is updated by the sensor every 10 Hz.

[0181] If the trajectory deviates from the predicted value, the dynamic potential field is regenerated.

[0182] Tools: Real-time running controller nodes, utilizing multi-threading for planning and sensing.

[0183] (5) Execution control

[0184] Path execution steps:

[0185] Adjust the robot's speed and direction according to the planned path.

[0186] The PID control algorithm is used to ensure path tracking accuracy.

[0187] Tools: Code for implementing PID control is embedded in STM32.

[0188] Error handling steps:

[0189] Emergency obstacle avoidance mode is triggered when an emergency obstacle is detected.

[0190] Prioritize finding the shortest path to bypass obstacles.

[0191] Specifically:

[0192] Global path planning (A* algorithm optimization): Global path planning is based on the A* algorithm and introduces a dynamic cost adjustment mechanism.

[0193]

[0194] in: :node The total cost. : From the starting point to the node The actual path cost. :node The heuristic cost to the target point is calculated using Euclidean distance: ; This is the dynamic cost, used to account for the impact of dynamic obstacles: ;in It is a node With dynamic obstacles distance, It is the obstacle weight.

[0195] Local path optimization (Dynamic Window Method, DWA): This method optimizes local paths using the Dynamic Window Method (DWA) with the goal of maximizing the overall score of the robot's motion.

[0196]

[0197] in: The set of speeds within the dynamic window. Alignment degree between current velocity direction and target direction: ;in It is the angle between the direction of velocity and the direction of the target; Minimum distance to the nearest obstacle in the velocity direction: ; It is the set of obstacle distances in the direction of velocity; : The magnitude of the current speed, representing the negative cost term.

[0198] A real-time update mechanism is used to update the path planning and calculate the update trigger conditions in real time during the robot's movement.

[0199]

[0200] in: Current robot position. The target position is predicted based on motion. If This triggers path replanning. This is the allowable deviation threshold.

[0201] By integrating dynamically updated paths, global paths and local optimization paths are weighted and combined:

[0202]

[0203] in: : Global path planning results. : Local path optimization results. : Dynamically adjust coefficients to adjust the weights of global and local paths based on the real-time environment.

[0204] Structural components: The campus litter-collecting robot in this embodiment consists of two parts: a mobile platform and a robotic arm. Its working principle is that the mobile platform carries the robotic arm and walks on the campus road. It identifies litter on the road through cameras and sensors. Then, the mobile platform moves to the location of the litter, stops, and the robotic arm picks up the litter and puts it into the litter bin on the side of the mobile platform.

[0205] The mobile platform adopts a rear-wheel drive and front-wheel steering mode. The rear wheels rotate by a stepper motor driving the chain drive, which in turn drives the shaft. The front wheels are driven by a steering mechanism designed according to the Ackermann steering geometry principle, which is driven by the rotation of a stepper motor.

[0206] To enable the trash-collecting robot to move freely on the campus grounds, the mobile platform adopts rear-wheel drive and front-wheel steering. A simplified structural diagram is shown below. Rear-wheel drive is used to control the front-wheel steering. The wheel rotation structure can be specifically divided into the following four parts:

[0207] 1. Two deep groove ball bearings are selected, which are axially positioned by the retaining ring and the motor housing shoulder, and radially positioned by the outer surface of the shaft.

[0208] 2. The motor is mounted on the motor frame, and then the motor and the motor frame are fixed together to the vehicle body. The positioning method is to rely on the edge of the motor frame for positioning.

[0209] 3. The wheels are positioned using bearings, and then connected to the coupling and steering shaft via the wheels. This process requires adjusting the concentricity of the motor shaft, and then the motor is secured to the frame through the pre-drilled mounting holes on the frame.

[0210] 4. The entire wheel consists of a wheel with an axle diameter and a wheel without an axle diameter. The two must be used together to form a complete wheel set. The wheel axle diameter serves as the axial positioning standard for the two wheels. The wheel is secured to the top plate via bolts on the outer side.

[0211] Overall design scheme of the robotic arm: Generally speaking, the main functions of the robotic arm are as follows: to improve working conditions, reduce the possibility of personal accidents, ensure the quality of the products produced, reduce the consumption of manpower and material resources, so as to facilitate efficient and orderly production, and improve the automation level of the production process. In this embodiment, the upper part of the campus garbage picking robot is the robotic arm. The function of the robotic arm is to pick up the garbage and put it into the garbage bin on the side of the mobile platform after the mobile platform moves to a suitable position, thus completing the task of picking up garbage. In order to complete the task, the robotic arm is designed with three degrees of freedom: (1) upper arm joint rotation; (2) forearm joint rotation; (3) gripper lifting.

[0212] Due to the large population flow on campus, a more flexible robotic arm with a large working range needs to be designed to facilitate its operation. Since the volume and weight of the waste are not too large, the designed robotic arm should be as simple and lightweight as possible.

[0213] Therefore, the basic parameters are selected as follows:

[0214] 1. Grabbing weight: Maximum 1kg;

[0215] 2. Number of degrees of freedom: 3;

[0216] 3. Structural type: Jointed;

[0217] 4. Upper arm length: 400mm;

[0218] 5. Forearm length: 320mm;

[0219] 6. Drive method: Stepper motor;

[0220] 7. Control method: Point-to-point program control;

[0221] 8. Gripping range of the gripper: up to 80mm;

[0222] 9. Upper arm and midarm joint rotation angle: 240°.

[0223] Because the robotic arm in this embodiment has three degrees of freedom, it can utilize a variety of structural combinations. This embodiment uses an articulated design, with each robotic arm connected by hinges, allowing it to rotate freely around the shoulder and elbow. Its features include high dexterity, a large working range, and strong practicality.

[0224] The drive system is a crucial element in robotic arm design. Based on the power source, common robotic arm drive systems mainly fall into four categories: hydraulic drive, pneumatic drive, electric motor drive, and mechanical drive. Common actuators include electric motor drives, hydraulic cylinder drives, and pneumatic drives. Among these, electric motor drive is the most common, characterized by high accuracy, strong controllability, and a wide speed range. Therefore, this robotic arm chooses an electric mechanism as its drive system, and the three most common electric motor drive methods are as follows:

[0225] 1. Stepper motor: High precision, strong controllability, can be stopped and started at any time, and has good positioning accuracy.

[0226] 2. DC servo motor: DC servo motors have good speed regulation characteristics, fast response, good stability, and are relatively easy to maintain and disassemble, with low cost.

[0227] 3. Direct drive: The motor is directly connected to the component it drives, without the need for a speed reduction device, which results in greater wear and tear on the motor.

[0228] After analysis, the robotic arm in this embodiment is ultimately driven by a stepper motor. This is because it has a simple structure, good positioning accuracy, is easy to disassemble, is inexpensive, and is highly efficient.

[0229] The configuration of the arm is a crucial aspect of the robotic arm's layout. Based on design requirements, to achieve the robotic arm's picking function, its working range needs to be sufficiently wide. Therefore, a column-type configuration was ultimately chosen, as it doesn't require a large space while still providing a large working range.

[0230] Based on the design requirements and the necessary technical requirements, the proposed structural design scheme for the robotic arm is as follows:

[0231] The robotic arm in this embodiment is required to have a simple structure and flexible movement, so the number of degrees of freedom is set to 4. These are (1) rotation of the upper arm joint; (2) rotation of the forearm joint; and (4) lifting and lowering of the gripper.

[0232] Working principle: A stepper motor at the upper arm joint drives a gear transmission to achieve the rotation of the upper arm; a stepper motor at the forearm joint drives a gear transmission to achieve the rotation of the forearm; the hand is raised and lowered by a single-axis drive. The main structure of the robotic hand consists of an upper arm, forearm, single-axis driver, and hand gripping mechanism. The base is made of cast aluminum as a whole, and the other joints are also made of cast aluminum, with the main shaft that drives the joint rotation and the corresponding bearings installed within it.

[0233] Material selection and weight estimation for the robotic arm: Since the robotic arm's function is to pick up trash, the grasping weight is very low. However, because the trash-collecting robot operates on campus, where there is high pedestrian traffic and complex road conditions, sufficient strength is required. Furthermore, to increase working speed, the material should be as lightweight as possible. The robotic arm material must not experience severe deformation or breakage under certain loads, thus requiring a certain level of strength. Therefore, mechanical materials such as cast aluminum or aluminum alloys should be selected. Moreover, since the arm is constantly in operation, its structure must be easy to control. Therefore, depending on the situation...

[0234] In summary, materials with sufficient strength and low density should be prioritized. After reviewing relevant materials, cast aluminum was ultimately chosen as the material for the robotic arm. It not only has sufficient strength but is also very lightweight, facilitating future disassembly and maintenance. The estimated weights of the upper arm, middle arm, and forearm are 2kg, 4kg, and 1kg, respectively.

[0235] Motor selection: Model 57BYGH56 motor is selected. Static torque is 1.2 Nm. This motor maintains sufficient and stable torque even as the speed increases.

[0236] The main purpose of this chain drive is to transmit the motor's rotation to the shaft, while eliminating the need to mount the motor on the shaft itself, thus reducing the load on the shaft. Therefore, the small sprocket speed is 180 r / min, and the large sprocket speed is also 180 r / min, with a center distance of no less than 250 mm. Both the large and small sprockets are selected with 17 teeth each. Chain pitch... p=12.7mm, 08 A Single-row chain, 67 links, number of sprocket teeth Z 1= Z 2=17, center distance a 1 = 292.1 mm, axial pressure F Q =93.75N.

[0237] Designing the forearm cross-section as a hollow column not only increases the bending coefficient but also reduces the total cross-sectional area, thereby reducing the overall weight and making it economical and lightweight.

[0238] Select ZL102 cast aluminum. Theoretical weight The forearm length is 320mm. ZL102 cast aluminum is a suitable material.

[0239] The boom is made of aluminum alloy because it is lightweight yet can withstand certain loads, simplifying the overall structure and reducing weight. ZL102 cast aluminum is a suitable choice.

[0240] The robotic arm doesn't require much driving force or high precision, so after consideration, a stepper motor with a simple structure and low cumulative error in output angle was ultimately chosen. The stepper motor selected is the 45BF005.

[0241] The lifting mechanism uses a single-axis drive, which is based on the principle of ball screw to control the lifting.

[0242] The gripper hand structure consists of two parts: fingers and a force transmission mechanism. The robotic hand uses a two-finger rotary mechanism. Since the workpiece is generally irregularly shaped or a cylindrical beverage bottle, the fingers are designed in a V-shape.

[0243] Design parameters for the clamp:

[0244] It uses a finger gripper, and the action is to grasp. Relax

[0245] (1) The maximum diameter of the workpiece to be gripped is 60mm;

[0246] (2) The maximum distance between the two hands when relaxed is 80mm;

[0247] (3) Grip tightly in 1 second, with a clamping speed of 10-20 mm / s;

[0248] (4) The maximum weight of the workpiece is 0.5 kg.

[0249] The above description of several specific embodiments further details the technical solution provided by the present invention in order to highlight the advantages and benefits of the technical solution provided by the present invention. However, the above-described specific embodiments are not intended to limit the present invention. Any reasonable modifications and improvements to the present invention, combinations of embodiments, and equivalent substitutions based on the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A method for planning the cleaning route of a campus garbage-collecting robot, characterized in that, include: The steps include constructing a time-series input of dynamic obstacles based on sensor data and generating the predicted reachable area of ​​the dynamic obstacles; The step of constructing a composite potential field including a static potential field, a dynamic potential field, and a gravitational field based on the predicted reachable area of ​​the dynamic obstacle; The steps include generating a global path based on the synthesized potential field and dynamically updating the global path based on real-time sensor data; Specifically: In the predicted reachable region of the dynamic obstacle, the future trajectory of the dynamic obstacle is predicted by a temporal convolutional network and a bidirectional GRU model; Based on the artificial potential field method, a composite potential field of static potential field, dynamic potential field and gravitational field is constructed. The static potential field calculates the repulsive force value through the geometric features of the obstacle and the Sigmoid function; the dynamic potential field is generated using the predicted region of the dynamic obstacle, and the weight decreases with the time step; the gravitational field generates the gradient gravity value based on the position of the target point; after the three potential fields are weighted and synthesized, the global path is dynamically updated based on real-time sensor data. The global path is generated using the A* algorithm, and the local path is optimized using the dynamic window method. The path is then dynamically updated based on real-time sensor data.

2. The method for planning the cleaning route of a campus garbage collection robot according to claim 1, characterized in that, The gravitational field is controlled by a Lyapunov function to generate a smooth gravitational gradient.

3. A campus trash-collecting robot, comprising a mobile platform and a robotic arm, wherein the mobile platform carries the robotic arm and enables the robot to move freely in the campus environment, and the robotic arm is used to pick up trash and place it onto a trash collection device on the mobile platform, characterized in that... The robot is used to implement the method of claim 1, including: The mobile platform includes a rear-wheel drive system and a front-wheel steering system. The rear-wheel drive system is driven by a stepper motor to rotate the rear wheels through chain transmission. The front-wheel steering system is based on the Ackermann steering geometry principle and is connected to the stepper motor through a coupling to achieve precise steering. The robotic arm is mounted on the top of the mobile platform and includes an upper arm, a forearm, and a gripper. The upper arm and forearm rotate through gear transmission, and the gripper is driven to lift and lower through a screw mechanism.

4. A computer storage medium for storing computer programs, characterized in that: When the computer program is read by the computer, the computer executes the method of claim 1.

5. A computer, including a processor and a storage medium, characterized in that, When the processor reads the computer program stored in the storage medium, the computer executes the method of claim 1.

6. A computer program product, as a computer program, is characterized by: When the computer program is executed, it implements the method of claim 1.

Images