Dynamic obstacle pre-judgment obstacle avoidance method and system for cleaning robot

Abstract

The invention belongs to the technical field of robot motion control, and discloses a dynamic obstacle pre-judgment obstacle avoidance method and system for a cleaning robot, and the method comprises the steps: obtaining the spatial distribution state information of obstacles in an operation environment of the cleaning robot and the motion state information of the cleaning robot; uniformly organizing the spatial distribution state information and the motion state information into a time sequence motion state set; on the basis of the time sequence motion state set, the motion expansion trend of the obstacle in a preset future time window is deduced, a motion expansion area is formed, spatial correlation analysis is conducted on the motion expansion area and a planned path of the cleaning robot, and a trend description result of the obstacle relative to the planned path is obtained; according to the trend description result, an obstacle risk potential energy function is constructed, the approaching intensity of the obstacle relative to the cleaning robot path in a preset future time window is represented, and the risk potential energy of the obstacle is obtained; the driving safety and the smooth path are guaranteed, and the cleaning efficiency and the environment adaptability of the cleaning robot are improved.

Description

Technical Field

[0001] This invention relates to the field of robot motion control technology, and more specifically, to a method and system for dynamic obstacle prediction and avoidance of a cleaning robot. Background Technology

[0002] Existing methods and systems for dynamic obstacle prediction and avoidance in cleaning robots mainly suffer from the following problems:

[0003] With the widespread application of intelligent cleaning robots in homes, offices, and public places, improving the robots' ability to autonomously avoid obstacles in dynamic environments has become a crucial technological aspect for ensuring safe operation and increasing cleaning efficiency. Existing methods and systems for dynamic obstacle avoidance in cleaning robots mainly rely on sensing the position and speed of obstacles for path planning and motion control, but they still have many technical shortcomings in practical applications.

[0004] Most existing dynamic obstacle avoidance methods predict the obstacle's future position based solely on its instantaneous position or velocity at the current moment. They typically employ the assumption of uniform velocity or simple linear extrapolation, failing to effectively characterize the trend of the obstacle's motion over time. When the obstacle accelerates, decelerates, or changes direction, the actual range of its future movement is easily underestimated, leading to delayed obstacle avoidance decisions or insufficient safety margins, thus posing a potential collision risk.

[0005] While existing technologies incorporate obstacle trajectory prediction, they largely rely on complex machine learning models or behavior recognition algorithms, which are highly dependent on training data and computational resources. In home or public environments, obstacles such as pedestrians and pets exhibit highly random movement behaviors, leading to insufficient stability and interpretability of prediction results, making reliable deployment on resource-constrained cleaning robot platforms difficult. Existing obstacle avoidance methods often rely on a single predicted point or trajectory for judgment, lacking a spatial representation of the uncertainty of obstacle future movement. This makes it difficult to form continuous and stable spatial constraints during path planning, easily leading to frequent replanning or path jitter, affecting cleaning efficiency and system reliability.

[0006] Existing technologies often assess risk based on instantaneous spatial distance or whether an obstacle enters a pre-defined safety zone. They lack a quantitative description of the obstacle's movement trend and its relative motion to the cleaning robot, making it difficult to accurately reflect the actual threat level of the obstacle to the planned path in the future. In dynamic scenarios, existing technologies typically only consider the obstacle's current position or absolute velocity, failing to model the obstacle's velocity in conjunction with the cleaning robot's own motion state. This leads to delayed or distorted risk assessments when the robot accelerates, decelerates, or turns. For dynamic obstacles with sudden movement behaviors, such as pedestrians and pets with sudden acceleration, changes of direction, or discontinuous trajectories, existing technologies lack explicit representation mechanisms for these behavioral abrupt changes, easily leading to underestimation of risk. This can result in collisions, frequent sudden stops, or path deviations, further impacting the safety and continuity of the cleaning task.

[0007] In view of this, the present invention proposes a dynamic obstacle prediction and obstacle avoidance method for cleaning robots to solve the above problems. Summary of the Invention

[0008] To overcome the aforementioned deficiencies of the prior art and to achieve the above objectives, the present invention provides the following technical solution: a dynamic obstacle prediction and avoidance method for a cleaning robot, comprising:

[0009] S1. Obtain the spatial distribution status information of obstacles in the operating environment of the cleaning robot and the motion status information of the cleaning robot, and organize the spatial distribution status information and motion status information into a temporal motion state set.

[0010] S2. Based on the temporal motion state set, the motion expansion trend of the obstacle within a preset future time window is inferred to form a motion expansion region. Spatial correlation analysis is then performed with the planned path of the cleaning robot to obtain the trend description result of the obstacle relative to the planned path.

[0011] S3. Based on the trend description results, construct the obstacle risk potential energy function to characterize the approach strength of the obstacle relative to the cleaning robot path within a preset future time window, and obtain the obstacle risk potential energy.

[0012] S4. When the risk potential energy of an obstacle exceeds a preset risk potential energy threshold, a dynamic safety corridor is generated in front of the cleaning robot, which dynamically adjusts with the obstacle; and the width of the dynamic safety corridor is adjusted according to the risk potential energy of the obstacle.

[0013] S5. Within the safe passage area defined by the dynamic safety corridor, the local path of the cleaning robot is replanned, and corresponding motion control commands are output to drive the cleaning robot to avoid dynamic obstacles in advance.

[0014] Preferably, the method for obtaining the spatial distribution state information and the motion state information of the cleaning robot includes:

[0015] During the operation of the cleaning robot, the environmental perception unit on the robot scans obstacles in the operating environment. At each sampling time, it acquires the spatial position information and occupancy boundary information of the obstacles relative to the cleaning robot, and combines the spatial position information and occupancy boundary information to represent the spatial distribution status information of the obstacles. The environmental perception unit includes a laser rangefinder, a vision sensor and a structured light sensor.

[0016] At the sampling moment corresponding to the acquisition of spatial distribution state information, the motion state information of the cleaning robot is synchronously acquired through the motion detection unit inside the cleaning robot. The motion detection unit includes an odometer and an inertial measurement unit, and the motion state information includes the pose information and motion speed information of the cleaning robot.

[0017] Preferably, the method for obtaining the time-series motion state set includes:

[0018] The spatial distribution state information and motion state information at the same sampling time are timestamped to construct a unified state description unit; different state description units are arranged and cached in sequence according to the sampling time order to form a temporal motion state set that represents the change of the relative motion relationship between the obstacle and the cleaning robot over time.

[0019] Preferably, the method for forming the motion extension region includes:

[0020] Based on the spatial position change relationship of obstacles in the time-series motion state set at different sampling times, the relative motion speed of the obstacles relative to the cleaning robot is determined by differentiating the spatial positions of the obstacles at adjacent sampling times, and the velocity change rate of the obstacles is estimated by differentiating the relative motion speed of the cleaning robot at adjacent time periods.

[0021] Within a preset future time window, by combining the current speed and rate of change of the obstacle, the maximum displacement range that the obstacle may produce within the future time window is evaluated, and the motion extension radius that characterizes the uncertainty of the obstacle's future motion is obtained.

[0022] Centered on the current spatial location of the obstacle, and based on the motion extension radius, under the unified motion reference frame adopted by the cleaning robot, the spatial range that the obstacle may reach within the future time window is constructed, and this spatial range is determined as the motion extension area of ​​the obstacle.

[0023] Preferably, the method for obtaining the trend description result of the obstacle relative to the planned path includes:

[0024] After forming the motion extension area, the cleaning robot generates a planned path based on the DWA algorithm during the cleaning task, and selects the planned path segment located within the preset future time window under a unified motion reference system.

[0025] The spatial relationship between the obstacle’s motion expansion area and the planned path segment within the preset future time window is calculated. The Euclidean distance from each path point on the planned path segment to each boundary position on the boundary of the motion expansion area is calculated as the spatial distance from the path point to the boundary of the motion expansion area. The minimum spatial distance from the path point to the boundary of the motion expansion area is taken within the planned path segment.

[0026] Based on the minimum spatial distance, the movement trend of the obstacle relative to the planned path within a preset future time window is determined; when the minimum spatial distance is less than the preset minimum spatial distance threshold, it is determined that the obstacle is approaching the planned path.

[0027] When the minimum spatial distance is greater than or equal to the preset minimum spatial distance threshold, it is determined that the obstacle is parallel or moving away from the planned path; and the determination result is used as the trend description result of the obstacle relative to the planned path.

[0028] Preferably, the method for obtaining the risk potential energy of an obstacle includes:

[0029] Based on the trend description results, for any perceived obstacle, an obstacle risk potential energy function is constructed to characterize the approach strength of the obstacle relative to the cleaning robot path within a preset future time window, so as to obtain the obstacle's risk potential energy.

[0030] Preferably, the method for generating a dynamic safety corridor that dynamically adjusts to obstacles includes:

[0031] When the risk potential energy of a perceived obstacle exceeds a preset risk potential energy threshold within a preset future time window, it is determined that the obstacle poses a dynamic intrusion risk to the cleaning robot's current planned path, and a dynamic safety corridor is generated in front of the cleaning robot.

[0032] Using the planned path generated by the cleaning robot at the current moment as a reference, a continuous dynamic safety corridor is constructed in front of the cleaning robot along the planned path direction under a unified motion reference system. The dynamic safety corridor covers the planned path segment that the cleaning robot will travel in the preset future time window in the longitudinal direction, and reserves preset safe passage space on both sides of the planned path in the transverse direction.

[0033] Preferably, the method for adjusting the width of the dynamic safety corridor includes:

[0034] Based on the risk potential energy of obstacles, the width of the dynamic safety corridor is adjusted. A preset basic safety width is used as the minimum width constraint of the dynamic safety corridor. The movement speed of obstacles is introduced to adaptively adjust the width of the dynamic safety corridor, so that the lateral width of the dynamic safety corridor is dynamically updated with the movement state corresponding to the risk potential energy of obstacles, thus keeping in line with the evolution trend of the risk potential energy of obstacles in the time dimension.

[0035] Preferably, the method for driving the cleaning robot to avoid dynamic obstacles in advance includes:

[0036] During the cleaning robot's execution of cleaning tasks and movement along the planned path, the local path of the cleaning robot within the safe passage area defined by the dynamic safety corridor is replanned in the future time window.

[0037] The path points of the cleaning robot's current planned path within a preset future time window are obtained and compared with the spatial constraints of the safe passage area defined by the dynamic safety corridor. Only the path points falling within the safe passage area defined by the dynamic safety corridor are retained. After the local path replanning is completed, motion control commands for the cleaning robot are generated simultaneously. The cleaning robot travels along the local path according to the motion control commands to achieve early avoidance of dynamic obstacles.

[0038] A dynamic obstacle prediction and avoidance system for a cleaning robot includes:

[0039] The motion state construction module is used to acquire spatial distribution information of obstacles in the operating environment of the cleaning robot and motion state information of the cleaning robot, and organize the spatial distribution information and motion state information into a temporal motion state set.

[0040] The obstacle behavior analysis module is used to infer the movement expansion trend of obstacles within a preset future time window based on a time-series motion state set, form a movement expansion region, and perform spatial correlation analysis with the cleaning robot's planned path to obtain the trend description results of obstacles relative to the planned path.

[0041] The obstacle intent prediction module is used to construct an obstacle risk potential energy function based on the trend description results, which characterizes the approach strength of the obstacle relative to the cleaning robot path within a preset future time window and obtains the obstacle's risk potential energy.

[0042] The safety corridor generation module is used to generate a dynamic safety corridor in front of the cleaning robot when the risk potential energy of an obstacle exceeds a preset risk potential energy threshold; and to adjust the width of the dynamic safety corridor according to the risk potential energy of the obstacle.

[0043] The obstacle avoidance path execution module is used to replan the local path of the cleaning robot within the safe passage area defined by the dynamic safety corridor, and output the corresponding motion control commands to drive the cleaning robot to avoid dynamic obstacles in advance.

[0044] Compared with the prior art, the present invention has the following beneficial effects:

[0045] This invention, by simultaneously introducing the relative velocity and rate of change of the obstacle, constructs a motion extension radius and a motion extension region, effectively covering the potential motion range of the obstacle within a future time window, avoiding the underprediction problem caused by simple uniform velocity assumptions. Obstacle motion trend inference is achieved based on geometric relationships and temporal difference methods, independent of training data or behavioral models, with strong interpretability in the calculation process, suitable for stable operation on cleaning robot platforms with limited computing power and power consumption. By constructing the motion extension region, the uncertainty of the obstacle's future motion is uniformly expressed as a clear spatial range, enabling the cleaning robot to directly judge spatial intrusion during path planning or safe corridor generation, reducing frequent replanning and path oscillations, and improving operational stability. Compared to reactive methods that only trigger obstacle avoidance when an obstacle approaches, this invention can identify potentially high-risk areas in advance, giving the cleaning robot predictive obstacle avoidance capabilities, and comprehensively improving operational safety and efficiency in complex dynamic environments.

[0046] By continuously quantifying the approach intensity of obstacles using a risk potential energy function, the system avoids discrete judgments based solely on distance or collision probability, enabling the cleaning robot to identify potentially high-risk obstacles in advance and achieve more proactive dynamic obstacle avoidance decisions. By employing the relative speed and distance between the obstacle and the cleaning robot as core parameters, the risk assessment results naturally adapt to changes in the robot's own motion state, improving the accuracy and stability of risk assessment under complex motion conditions such as acceleration, deceleration, and turning. By introducing behavioral mutation indicators to comprehensively quantify the magnitude and direction of changes in obstacle motion speed, the system can effectively identify sudden behavioral characteristics of obstacles, reducing the risk of misjudgment due to the uncertainty of obstacle motion, thereby improving obstacle avoidance safety. Attached Figure Description

[0047] Figure 1 This is a schematic diagram of a dynamic obstacle prediction and obstacle avoidance method for a cleaning robot according to the present invention;

[0048] Figure 2 This is a schematic diagram of the structure of a dynamic obstacle prediction and avoidance system for a cleaning robot according to the present invention;

[0049] Figure 3 This is a schematic diagram of the dynamic safety corridor provided by the present invention. Detailed Implementation

[0050] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0051] Example 1: Please refer to Figure 1 As shown, this embodiment provides a dynamic obstacle prediction and avoidance method for a cleaning robot, specifically including the following steps:

[0052] S1. Obtain the spatial distribution status information of obstacles in the operating environment of the cleaning robot and the motion status information of the cleaning robot, and organize the spatial distribution status information and motion status information into a temporal motion state set.

[0053] S2. Based on the temporal motion state set, the motion expansion trend of the obstacle within a preset future time window is inferred to form a motion expansion region. Spatial correlation analysis is then performed with the planned path of the cleaning robot to obtain the trend description result of the obstacle relative to the planned path.

[0054] S3. Based on the trend description results, construct the obstacle risk potential energy function to characterize the approach strength of the obstacle relative to the cleaning robot path within a preset future time window, and obtain the obstacle risk potential energy.

[0055] S4. When the risk potential energy of an obstacle exceeds a preset risk potential energy threshold, a dynamic safety corridor is generated in front of the cleaning robot, which dynamically adjusts with the obstacle; and the width of the dynamic safety corridor is adjusted according to the risk potential energy of the obstacle.

[0056] S5. Within the safe passage area defined by the dynamic safety corridor, the local path of the cleaning robot is replanned, and corresponding motion control commands are output to drive the cleaning robot to avoid dynamic obstacles in advance.

[0057] Methods for acquiring spatial distribution state information and motion state information of cleaning robots include:

[0058] During the operation of the cleaning robot, the environmental perception unit on the robot scans obstacles in the operating environment. At each sampling time, it acquires the spatial position information and occupancy boundary information of the obstacles relative to the cleaning robot, and combines the spatial position information and occupancy boundary information to represent the spatial distribution status information of the obstacles. The environmental perception unit includes a laser rangefinder, a vision sensor and a structured light sensor.

[0059] It should be noted that for laser rangefinders, during operation, they emit laser beams along a preset angle range and receive reflected signals. The distance measurement values ​​in each direction are calculated based on the time difference between laser emission and reception. By combining the distance measurement values ​​with the laser scanning angle, a set of spatial points around the cleaning robot can be obtained, thus determining the spatial distribution of the obstacles relative to the robot. Based on this spatial point set, clustering or contour extraction of adjacent points can further determine the equivalent spatial boundary of the obstacle. For vision sensors, based on acquired images of the operating environment, they analyze spatially continuous regions in the images to identify corresponding obstacle regions. Combining image imaging parameters and depth or multi-view information, the spatial position of the obstacle relative to the cleaning robot is estimated. Simultaneously, based on the projection range of the obstacle region in the image and combined with spatial scale conversion relationships, the contour of the obstacle's actual spatial boundary is obtained.

[0060] Structured light sensors project light patterns with known structural features into the environment and collect data on the deformation of these patterns on obstacle surfaces to calculate depth information. Based on this depth information, a three-dimensional point cloud of obstacles around the cleaning robot can be directly constructed, thereby determining the spatial position of the obstacles relative to the cleaning robot and their corresponding occupancy boundaries.

[0061] At the sampling moment corresponding to the acquisition of spatial distribution state information, the motion state information of the cleaning robot is synchronously acquired through the motion detection unit inside the cleaning robot. The motion detection unit includes an odometer and an inertial measurement unit, and the motion state information includes the pose information and motion speed information of the cleaning robot.

[0062] It should be noted that the odometer calculates the robot's displacement in a plane by detecting the rotation of its drive wheels and combining this with wheel diameter parameters and a kinematic model. The inertial measurement unit (IMU) calculates the robot's attitude changes and motion trends by collecting acceleration and angular velocity information during operation. By fusing the displacement information output by the odometer with the attitude and acceleration information output by the IMU, the robot's pose and velocity information at the current sampling moment are obtained, thus forming the robot's motion state information.

[0063] Methods for obtaining temporal motion state sets include:

[0064] The spatial distribution state information and motion state information at the same sampling time are timestamped to construct a unified state description unit; different state description units are arranged and cached in sequence according to the sampling time order to form a temporal motion state set that represents the change of the relative motion relationship between the obstacle and the cleaning robot over time.

[0065] Methods for forming motion extension regions include:

[0066] Based on the spatial position change relationship of obstacles in the time-series motion state set at different sampling times, the relative motion speed of the obstacles relative to the cleaning robot is determined by differentiating the spatial positions of the obstacles at adjacent sampling times, and the velocity change rate of the obstacles is estimated by differentiating the relative motion speed of the cleaning robot at adjacent time periods.

[0067] It should be noted that at two adjacent sampling times, the spatial position of the obstacle in the cleaning robot's reference frame is recorded respectively. By differential processing of the obstacle's spatial position at adjacent sampling times, and combining it with the corresponding sampling time interval, the average relative motion velocity of the obstacle within that time interval is obtained, which is used to characterize the relative motion velocity of the obstacle relative to the cleaning robot. At multiple consecutive sampling times, a time series analysis of the relative motion velocity is further performed. Specifically, by differential processing of the relative motion velocities calculated in adjacent time periods, and combining it with the corresponding time interval, the change in the relative motion velocity of the obstacle is obtained, which is used to characterize the rate of change of the obstacle's velocity.

[0068] Within a preset future time window, by combining the current speed and rate of change of the obstacle, the maximum displacement range that the obstacle may produce within the future time window is evaluated, and the motion extension radius that characterizes the uncertainty of the obstacle's future motion is obtained.

[0069] The radius of motion extension is: ;in, Indicates obstacles In time Starting from, the duration is The radius of motion expansion within the preset future time window; This indicates the magnitude of the relative velocity of the obstacle with respect to the cleaning robot; Indicates the intensity of the trend of change in the obstacle's speed; Indicates obstacles At any moment The velocity vector relative to the cleaning robot; Indicates obstacles At any moment The rate of change of velocity; Indicates a preset future time window; An index representing a time point; This refers to a specific obstacle object that is perceived and participates in obstacle avoidance analysis.

[0070] Centered on the current spatial location of the obstacle, and based on the motion extension radius, under the unified motion reference frame adopted by the cleaning robot, the spatial range that the obstacle may reach within the future time window is constructed, and this spatial range is determined as the motion extension area of ​​the obstacle.

[0071] The motion extension area is: ;in, Indicates obstacles In the preset future time window The motion extension area within; Represents any point in space within a moving reference frame; Represents a spatial location point With obstacles at any time Current location Spatial distance between them; Indicates obstacles At any moment The spatial position vector relative to the cleaning robot; This indicates a distance constraint relationship, used to limit all spatial points whose distance from the current position of the obstacle does not exceed the radius of motion expansion to be considered as areas that the obstacle may reach in the future.

[0072] This solution addresses the following problems in existing technologies: Most current dynamic obstacle avoidance methods only predict the obstacle's instantaneous position or velocity at the current moment, typically using a uniform velocity assumption or simple linear extrapolation to infer the obstacle's future position, failing to effectively characterize the trend of the obstacle's motion state changing over time. When the obstacle accelerates, decelerates, or adjusts its direction of motion, it is easy to underestimate the obstacle's actual range of motion in the future, resulting in delayed obstacle avoidance decisions or insufficient safety margins.

[0073] While existing technologies incorporate obstacle trajectory prediction, they largely rely on complex machine learning models or behavior recognition algorithms. This not only places a heavy burden on training data and computational resources, but also results in insufficient stability and interpretability when facing highly random obstacles such as pedestrians and pets in home or public environments, making reliable deployment on resource-constrained cleaning robot platforms difficult. Existing obstacle avoidance methods often rely on predicted points or trajectories for judgment, lacking a spatial representation of the uncertainties in the future movement of obstacles. This makes it difficult to form continuous and stable spatial constraints during path planning, easily leading to frequent replanning or path jitter problems.

[0074] The advantages over existing technologies are as follows: By simultaneously introducing the relative velocity and rate of change of the obstacle, and constructing the motion extension radius and motion extension region, the potential motion range of the obstacle within the future time window can be effectively covered, avoiding the underprediction problem caused by the simple uniform velocity assumption. Obstacle motion trend inference is achieved based on geometric relationships and temporal difference methods, without relying on training data or behavioral models. The calculation process is highly interpretable and suitable for stable operation on cleaning robot platforms with limited computing power and power consumption. By constructing the motion extension region, the uncertainty of the obstacle's future motion is uniformly expressed as a clear spatial range, enabling the cleaning robot to directly judge spatial intrusion when planning paths or generating safe corridors, reducing frequent replanning and path oscillations, and improving operational stability. Compared to reactive methods that only trigger obstacle avoidance when an obstacle approaches, this invention can identify potentially high-risk areas in advance, enabling the cleaning robot to have predictive obstacle avoidance capabilities, thus improving the overall operational safety and efficiency in complex dynamic environments.

[0075] Methods for obtaining trend descriptions of obstacles relative to the planned path include:

[0076] After forming the motion extension area, the cleaning robot generates a planned path based on the DWA algorithm during the cleaning task, and selects the planned path segment located within the preset future time window under a unified motion reference system.

[0077] The spatial relationship between the obstacle’s motion expansion area and the planned path segment within the preset future time window is calculated. The Euclidean distance from each path point on the planned path segment to each boundary position on the boundary of the motion expansion area is calculated as the spatial distance from the path point to the boundary of the motion expansion area. The minimum spatial distance from the path point to the boundary of the motion expansion area is taken within the planned path segment.

[0078] Based on the minimum spatial distance, the movement trend of the obstacle relative to the planned path within a preset future time window is determined; when the minimum spatial distance is less than the preset minimum spatial distance threshold, it is determined that the obstacle is approaching the planned path.

[0079] When the minimum spatial distance is greater than or equal to the preset minimum spatial distance threshold, it is determined that the obstacle is parallel or moving away from the planned path; and the determination result is used as the trend description result of the obstacle relative to the planned path.

[0080] Methods for obtaining the risk potential energy of obstacles include:

[0081] Based on the trend description results, for any perceived obstacle, an obstacle risk potential energy function is constructed to characterize the approach strength of the obstacle relative to the cleaning robot path within a preset future time window, so as to obtain the obstacle's risk potential energy.

[0082] The obstacle risk potential energy function is: ;in, Indicates obstacles At any moment The risk potential energy relative to cleaning robots; Indicates cleaning robot At any moment The velocity vector of the robot's motion is obtained by the robot's own motion detection unit. Indicates obstacles With cleaning robots At any moment The magnitude of the relative velocity; Indicates cleaning robot and obstacles At any moment Spatial distance between them; This represents the distance stability factor used to prevent numerical divergence or computational instability when the relative distance is small. Indicates obstacles At any moment The behavioral mutation index is obtained by normalizing the magnitude of changes in the speed and direction of obstacle movement, and then weighting and fusing them according to preset weights.

[0083] This solution addresses the following problems with existing technologies: Existing technologies often judge obstacles based on instantaneous spatial distance or whether the obstacle has entered a safe zone, lacking a quantitative description of the obstacle's movement trend and relative motion relationship. This makes it difficult to accurately reflect the actual threat level of the obstacle to the planned path in the future. In dynamic scenarios, existing technologies typically only consider the obstacle's current position or absolute velocity, without uniformly modeling the obstacle's velocity with the cleaning robot's own motion state. This leads to lagging or distorted risk assessment results when the robot accelerates, decelerates, or turns. For dynamic obstacles with sudden movement behaviors (such as pedestrians or pets with sudden acceleration, change of direction, or discontinuous trajectories), existing technologies lack explicit representation mechanisms for behavioral abrupt changes, easily leading to underestimation of risk and resulting in collisions or frequent sudden stops.

[0084] Compared to existing technologies, the advantages are as follows: By continuously quantifying the approach intensity of obstacles through a risk potential energy function, the discrete judgment method based solely on distance or collision status is avoided. This allows the cleaning robot to identify potentially high-risk obstacles in advance, enabling more proactive dynamic obstacle avoidance decisions. By using the relative speed and distance between the obstacle and the cleaning robot as core parameters, the risk assessment results are naturally adapted to changes in the robot's own motion state, improving the accuracy and stability of risk assessment under complex motion conditions such as acceleration, deceleration, and turning. By introducing behavioral mutation indicators to comprehensively quantify the magnitude and direction of changes in obstacle motion speed, the system can effectively identify sudden behavioral characteristics of obstacles, reducing the risk of misjudgment due to the uncertainty of obstacle motion, thereby improving obstacle avoidance safety.

[0085] Methods for generating dynamic safety corridors that adjust dynamically to obstacles include:

[0086] When the risk potential energy of a perceived obstacle exceeds a preset risk potential energy threshold within a preset future time window, it is determined that the obstacle poses a dynamic intrusion risk to the cleaning robot's current planned path, and a dynamic safety corridor (DSC) is generated in front of the cleaning robot.

[0087] Using the planned path generated by the cleaning robot at the current moment as a reference, a continuous dynamic safety corridor is constructed in front of the cleaning robot along the planned path direction under a unified motion reference system. The dynamic safety corridor covers the planned path segment that the cleaning robot will travel in the preset future time window in the longitudinal direction, and reserves preset safe passage space on both sides of the planned path in the transverse direction.

[0088] Methods for adjusting the width of dynamic safety corridors include:

[0089] Based on the risk potential energy of obstacles, the width of the dynamic safety corridor is adjusted. A preset basic safety width is used as the minimum width constraint of the dynamic safety corridor. The movement speed of obstacles is introduced to adaptively adjust the width of the dynamic safety corridor, so that the lateral width of the dynamic safety corridor is dynamically updated with the movement state corresponding to the risk potential energy of obstacles, thus keeping in line with the evolution trend of the risk potential energy of obstacles in the time dimension.

[0090] The width of the dynamic safety corridor is adaptively adjusted as follows: ;in, Indicates time The real-time width of the dynamic safety corridor; Indicates the preset basic safety width; This represents the speed adjustment coefficient, used to control the impact of obstacle movement speed on the extent of safe corridor expansion.

[0091] Methods for driving cleaning robots to avoid dynamic obstacles in advance include:

[0092] During the cleaning robot's execution of cleaning tasks and movement along the planned path, the local path of the cleaning robot within the safe passage area defined by the dynamic safety corridor is replanned in the future time window.

[0093] Obtain the path points of the cleaning robot's current planned path within a preset future time window, and compare them with the spatial constraints of the safe passage area defined by the dynamic safety corridor, retaining only the path points that fall within the safe passage area defined by the dynamic safety corridor;

[0094] After the local path replanning is completed, motion control commands for the cleaning robot are generated synchronously. These commands include speed commands, acceleration commands, steering angle commands, and attitude control commands. The motion control commands directly depend on the latest local path geometry information, ensuring that the robot smoothly and promptly avoids dynamic obstacles along the safe passage area. The cleaning robot travels along the local path according to the motion control commands, thus achieving early avoidance of dynamic obstacles.

[0095] The preset risk potential energy threshold is set by staff based on historical data analysis results. This historical analysis process includes the system collecting multiple risk potential energies and calculating their average value as a reference to obtain the preset risk potential energy threshold. Similarly, the preset minimum spatial distance threshold is also set by staff based on the system's historical operating data and the specific application scenario requirements, and can be adjusted by staff during system operation according to the actual situation.

[0096] This embodiment, by simultaneously introducing the relative velocity and rate of change of the obstacle, constructs a motion extension radius and a motion extension region, effectively covering the potential motion range of the obstacle within the future time window, avoiding the underprediction problem caused by the simple uniform velocity assumption. Obstacle motion trend inference is achieved based on geometric relationships and temporal difference methods, without relying on training data or behavioral models. The calculation process is highly interpretable and suitable for stable operation on cleaning robot platforms with limited computing power and power consumption. By constructing the motion extension region, the uncertainty of the obstacle's future motion is uniformly expressed as a clear spatial range, enabling the cleaning robot to directly judge spatial intrusion when planning paths or generating safe corridors, reducing frequent replanning and path oscillations, and improving operational stability. Compared to reactive methods that only trigger obstacle avoidance when an obstacle approaches, this invention can identify potentially high-risk areas in advance, giving the cleaning robot predictive obstacle avoidance capabilities, and comprehensively improving operational safety and efficiency in complex dynamic environments.

[0097] By continuously quantifying the approach intensity of obstacles using a risk potential energy function, the system avoids discrete judgments based solely on distance or collision probability, enabling the cleaning robot to identify potentially high-risk obstacles in advance and achieve more proactive dynamic obstacle avoidance decisions. By employing the relative speed and distance between the obstacle and the cleaning robot as core parameters, the risk assessment results naturally adapt to changes in the robot's own motion state, improving the accuracy and stability of risk assessment under complex motion conditions such as acceleration, deceleration, and turning. By introducing behavioral mutation indicators to comprehensively quantify the magnitude and direction of changes in obstacle motion speed, the system can effectively identify sudden behavioral characteristics of obstacles, reducing the risk of misjudgment due to the uncertainty of obstacle motion, thereby improving obstacle avoidance safety.

[0098] Example 2: Please refer to Figure 2 As shown, parts not described in detail in this embodiment are described in Embodiment 1. A dynamic obstacle prediction and avoidance system for a cleaning robot is provided, including:

[0099] The motion state construction module is used to acquire spatial distribution information of obstacles in the operating environment of the cleaning robot and motion state information of the cleaning robot, and organize the spatial distribution information and motion state information into a temporal motion state set.

[0100] The obstacle behavior analysis module is used to infer the movement expansion trend of obstacles within a preset future time window based on a time-series motion state set, form a movement expansion region, and perform spatial correlation analysis with the cleaning robot's planned path to obtain the trend description results of obstacles relative to the planned path.

[0101] The obstacle intent prediction module is used to construct an obstacle risk potential energy function based on the trend description results, which characterizes the approach strength of the obstacle relative to the cleaning robot path within a preset future time window and obtains the obstacle's risk potential energy.

[0102] The safety corridor generation module is used to generate a dynamic safety corridor in front of the cleaning robot when the risk potential energy of an obstacle exceeds a preset risk potential energy threshold; and to adjust the width of the dynamic safety corridor according to the risk potential energy of the obstacle.

[0103] The obstacle avoidance path execution module is used to replan the local path of the cleaning robot within the safe passage area defined by the dynamic safety corridor, and output the corresponding motion control commands to drive the cleaning robot to avoid dynamic obstacles in advance.

[0104] Since the electronic device described in this embodiment is the electronic device used to implement the dynamic obstacle prediction and avoidance method and system for a cleaning robot based on the embodiments of this application, those skilled in the art can understand the specific implementation methods and various variations of the electronic device in this embodiment based on the dynamic obstacle prediction and avoidance method and system for a cleaning robot described in the embodiments of this application. Therefore, how the electronic device implements the method in the embodiments of this application will not be described in detail here. As long as those skilled in the art implement the electronic device used in the dynamic obstacle prediction and avoidance method and system for a cleaning robot based on the embodiments of this application, it falls within the scope of protection of this application.

[0105] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0106] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principles of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for dynamic obstacle prediction and avoidance in a cleaning robot, characterized in that, include: S1. Obtain the spatial distribution status information of obstacles in the operating environment of the cleaning robot and the motion status information of the cleaning robot, and organize the spatial distribution status information and motion status information into a temporal motion state set. S2. Based on the temporal motion state set, the motion expansion trend of the obstacle within a preset future time window is inferred to form a motion expansion region. Spatial correlation analysis is then performed with the planned path of the cleaning robot to obtain the trend description result of the obstacle relative to the planned path. S3. Based on the trend description results, construct the obstacle risk potential energy function to characterize the approach strength of the obstacle relative to the cleaning robot path within a preset future time window, and obtain the obstacle risk potential energy. S4. When the risk potential energy of an obstacle exceeds the preset risk potential energy threshold, a dynamic safety corridor is generated in front of the cleaning robot, which is dynamically adjusted according to the obstacle. The width of the dynamic safety corridor is adjusted based on the potential energy of the obstacles. S5. Within the safe passage area defined by the dynamic safety corridor, the local path of the cleaning robot is replanned, and corresponding motion control commands are output to drive the cleaning robot to avoid dynamic obstacles in advance.

2. The method for dynamic obstacle prediction and avoidance of a cleaning robot according to claim 1, characterized in that, The methods for obtaining the spatial distribution state information and the motion state information of the cleaning robot include: During the operation of the cleaning robot, the environmental perception unit on the robot scans obstacles in the operating environment. At each sampling time, it acquires the spatial position information and occupancy boundary information of the obstacles relative to the cleaning robot, and combines the spatial position information and occupancy boundary information to represent the spatial distribution status information of the obstacles. The environmental perception unit includes a laser rangefinder, a vision sensor and a structured light sensor. At the sampling moment corresponding to the acquisition of spatial distribution state information, the motion state information of the cleaning robot is synchronously acquired through the motion detection unit inside the cleaning robot. The motion detection unit includes an odometer and an inertial measurement unit, and the motion state information includes the pose information and motion speed information of the cleaning robot.

3. The method for dynamic obstacle prediction and avoidance of a cleaning robot according to claim 2, characterized in that, The method for obtaining the temporal motion state set includes: The spatial distribution state information and motion state information at the same sampling time are timestamped to construct a unified state description unit; different state description units are arranged and cached in sequence according to the sampling time order to form a temporal motion state set that represents the change of the relative motion relationship between the obstacle and the cleaning robot over time.

4. The method for dynamic obstacle prediction and avoidance of a cleaning robot according to claim 3, characterized in that, The method for forming the motion extension region includes: Based on the spatial position change relationship of obstacles in the time-series motion state set at different sampling times, the relative motion speed of the obstacles relative to the cleaning robot is determined by differentiating the spatial positions of the obstacles at adjacent sampling times, and the velocity change rate of the obstacles is estimated by differentiating the relative motion speed of the cleaning robot at adjacent time periods. Within a preset future time window, by combining the current speed and rate of change of the obstacle, the maximum displacement range that the obstacle may produce within the future time window is evaluated, and the motion extension radius that characterizes the uncertainty of the obstacle's future motion is obtained. Centered on the current spatial location of the obstacle, and based on the motion extension radius, under the unified motion reference frame adopted by the cleaning robot, the spatial range that the obstacle may reach within the future time window is constructed, and this spatial range is determined as the motion extension area of ​​the obstacle.

5. The method for dynamic obstacle prediction and avoidance of a cleaning robot according to claim 4, characterized in that, The method for obtaining the trend description of obstacles relative to the planned path includes: After forming the motion extension area, the cleaning robot generates a planned path based on the DWA algorithm during the cleaning task, and selects the planned path segment located within the preset future time window under a unified motion reference system. The spatial relationship between the obstacle’s motion expansion area and the planned path segment within the preset future time window is calculated. The Euclidean distance from each path point on the planned path segment to each boundary position on the boundary of the motion expansion area is calculated as the spatial distance from the path point to the boundary of the motion expansion area. The minimum spatial distance from the path point to the boundary of the motion expansion area is taken within the planned path segment. Based on the minimum spatial distance, the movement trend of the obstacle relative to the planned path within a preset future time window is determined; when the minimum spatial distance is less than the preset minimum spatial distance threshold, it is determined that the obstacle is approaching the planned path. When the minimum spatial distance is greater than or equal to the preset minimum spatial distance threshold, it is determined that the obstacle is parallel or moving away from the planned path; and the determination result is used as the trend description result of the obstacle relative to the planned path.

6. The method for dynamic obstacle prediction and avoidance of a cleaning robot according to claim 5, characterized in that, The method for obtaining the risk potential energy of an obstacle includes: Based on the trend description results, for any perceived obstacle, an obstacle risk potential energy function is constructed to characterize the approach strength of the obstacle relative to the cleaning robot path within a preset future time window, so as to obtain the obstacle's risk potential energy.

7. The method for dynamic obstacle prediction and avoidance of a cleaning robot according to claim 6, characterized in that, The method for generating a dynamic safety corridor that dynamically adjusts to obstacles includes: When the risk potential energy of a perceived obstacle exceeds a preset risk potential energy threshold within a preset future time window, it is determined that the obstacle poses a dynamic intrusion risk to the cleaning robot's current planned path, and a dynamic safety corridor is generated in front of the cleaning robot. Using the planned path generated by the cleaning robot at the current moment as a reference, a continuous dynamic safety corridor is constructed in front of the cleaning robot along the planned path direction under a unified motion reference system. The dynamic safety corridor covers the planned path segment that the cleaning robot will travel in the preset future time window in the longitudinal direction, and reserves preset safe passage space on both sides of the planned path in the transverse direction.

8. A dynamic obstacle prediction and avoidance method for a cleaning robot according to claim 7, characterized in that, The method for adjusting the width of the dynamic safety corridor includes: Based on the risk potential energy of obstacles, the width of the dynamic safety corridor is adjusted. A preset basic safety width is used as the minimum width constraint of the dynamic safety corridor. The movement speed of obstacles is introduced to adaptively adjust the width of the dynamic safety corridor, so that the lateral width of the dynamic safety corridor is dynamically updated with the movement state corresponding to the risk potential energy of obstacles, thus keeping in line with the evolution trend of the risk potential energy of obstacles in the time dimension.

9. A dynamic obstacle prediction and avoidance method for a cleaning robot according to claim 8, characterized in that, The method for driving the cleaning robot to avoid dynamic obstacles in advance includes: During the cleaning robot's execution of cleaning tasks and movement along the planned path, the local path of the cleaning robot within the safe passage area defined by the dynamic safety corridor is replanned in the future time window. The path points of the cleaning robot's current planned path within a preset future time window are obtained and compared with the spatial constraints of the safe passage area defined by the dynamic safety corridor. Only the path points falling within the safe passage area defined by the dynamic safety corridor are retained. After the local path replanning is completed, motion control commands for the cleaning robot are generated simultaneously. The cleaning robot travels along the local path according to the motion control commands to achieve early avoidance of dynamic obstacles.

10. A dynamic obstacle prediction and avoidance system for a cleaning robot, used to implement the dynamic obstacle prediction and avoidance method for a cleaning robot according to any one of claims 1 to 9, characterized in that, include: The motion state construction module is used to acquire spatial distribution information of obstacles in the operating environment of the cleaning robot and motion state information of the cleaning robot, and organize the spatial distribution information and motion state information into a temporal motion state set. The obstacle behavior analysis module is used to infer the movement expansion trend of obstacles within a preset future time window based on a time-series motion state set, form a movement expansion region, and perform spatial correlation analysis with the cleaning robot's planned path to obtain the trend description results of obstacles relative to the planned path. The obstacle intent prediction module is used to construct an obstacle risk potential energy function based on the trend description results, which characterizes the approach strength of the obstacle relative to the cleaning robot path within a preset future time window and obtains the obstacle's risk potential energy. The safety corridor generation module is used to generate a dynamic safety corridor in front of the cleaning robot when the risk potential energy of an obstacle exceeds a preset risk potential energy threshold. The width of the dynamic safety corridor is adjusted based on the potential energy of the obstacles. The obstacle avoidance path execution module is used to replan the local path of the cleaning robot within the safe passage area defined by the dynamic safety corridor, and output the corresponding motion control commands to drive the cleaning robot to avoid dynamic obstacles in advance.

Images