A mobile air purifier path planning and obstacle avoidance method and system, and a mobile air purifier

Abstract

The present application relates to a kind of mobile air purifier path planning and obstacle avoidance method, system and mobile air purifier, wherein the method comprises: step 1, multi-sensor group data acquisition;Step 2, space-time calibration;Step 3, fusion modeling;Step 4, passage judgment;Step 5, global planning;Step 6, local obstacle avoidance;Step 7, task linkage;Step 8, execution control;Step 9, loop execution.The present application adopts three-dimensional path planning scheme based on laser radar and multi-sensor fusion, by constructing dynamic passable area model, combined with hierarchical obstacle avoidance strategy and task priority scheduling, can realize safe, efficient autonomous navigation and accurate purification.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of mobile air purifiers, and particularly relates to a mobile air purifier path planning and obstacle avoidance method and system and a mobile air purifier. BACKGROUND

[0002] The existing mobile air purifier path planning and obstacle avoidance method has the following technical defects:

[0003] Two-dimensional perception limitation: the traditional mobile air purifier relies on two-dimensional laser radar or infrared sensors, and can only obtain horizontal plane obstacle information, and cannot judge the passability in the vertical height direction (such as the bottom of a table and chair, a low passage), resulting in the risk of being trapped or colliding (occurrence rate of about 25%).

[0004] Path planning is rough: the existing algorithm (such as fixed route planning) lacks dynamic adaptability to the environment, and the purification efficiency is significantly reduced when encountering complex obstacles (the purification time is prolonged by 30%-40% for the same area), and the purification task priority is not combined, and the hot pollution area cannot be efficiently responded.

[0005] Insufficient multi-sensor fusion: although there are some multi-sensor solutions, a special fusion algorithm is not designed for the low-height body characteristics (height is usually ≤50cm) of the air purifier, and the safety margin setting lacks engineering verification, and the success rate of passing in the actual scene is only 68%. SUMMARY

[0006] The purpose of the application is to provide a mobile air purifier path planning and obstacle avoidance method and system and a mobile air purifier, which adopts a three-dimensional path planning scheme based on laser radar and multi-sensor fusion for the navigation and purification needs of the mobile air purifier in a complex home environment, constructs a dynamic passable area model, combines a hierarchical obstacle avoidance strategy and task priority scheduling, and realizes safe and efficient autonomous navigation and accurate purification.

[0007] The application provides a mobile air purifier path planning and obstacle avoidance method, which comprises the following steps:

[0008] Step 1, multi-sensor group data acquisition:

[0009] The distance and angle data of the horizontal plane obstacle are obtained by the two-dimensional laser radar, the height and three-dimensional contour data of the obstacle are collected in real time by the ToF sensor and RGB-D camera, and the pollution concentration data of PM2.5 and formaldehyde in the current environment are collected by the air quality sensor;

[0010] Step 2, space-time calibration:

[0011] The data collected in step 1 is time and space synchronized to ensure consistency in time and space, with an error of less than 5ms;

[0012] Step 3, fusion modeling:

[0013] The extended Kalman filter algorithm is used to fuse the data of two-dimensional laser radar and ToF sensor to generate a three-dimensional voxel map with a voxel resolution of 10cm×10cm×10cm, and each voxel contains an occupancy state and a height value;

[0014] Step 4, passage judgment:

[0015] According to the safety margin set by the size of the air purifier, mark the area with height <H+h or horizontal size <L+l / W+w as an impassable area; wherein L, W, H are the length, width and height of the air purifier respectively; h is the height safety margin, l, w are the horizontal safety margins corresponding to the length and width of the air purifier respectively;

[0016] Step 5, global planning:

[0017] A* algorithm is used for global path planning:

[0018] According to the complexity of the environment, the heuristic function is dynamically adjusted, and in simple environment, Manhattan distance is used to accelerate search; in complex environment, Euclidean distance and dynamic weight combination are switched;

[0019] Avoid impassable areas when planning paths, and prefer paths through pollution hotspots;

[0020] Step 6, local obstacle avoidance:

[0021] Velocity sample generation: based on the current velocity, 50 groups of velocity samples are generated within the range of current velocity ±20%;

[0022] Trajectory simulation: simulate the trajectory of each group of velocity samples for the next 2s;

[0023] Height filtering: remove samples in the trajectory that pass through the height <H+h area;

[0024] Weighted scoring: score the remaining samples according to safety 40%, smoothness 30%, and target approach 30%;

[0025] Select the optimal trajectory: according to the weighted score result, select the optimal trajectory as the current local obstacle avoidance path;

[0026] Step 7, task linkage:

[0027] Based on the air quality sensor data, the DBSCAN clustering algorithm is used to identify the area with concentration greater than the threshold, and a hot area priority queue is generated; the threshold is PM2.5>75μg / m 3 Or formaldehyde>0.08mg / m 3 ;

[0028] Determine whether a pollution hot area is detected:

[0029] If yes: interrupt the current path, insert the hot area purification task, and use the dynamic window method to plan the shortest collision-free path to the hot area;

[0030] If no: continue to execute the current path planning;

[0031] Step 8, control execution:

[0032] The path planning result is converted into a motor control signal of the wheeled chassis through a PID control algorithm, and the air purifier is driven to move according to the planned path;

[0033] Step 9, loop execution:

[0034] The air purifier continues to return to step 1 for data collection during movement, constantly updates environmental information, realizes real-time path planning and obstacle avoidance adjustment, and forms a closed-loop control process.

[0035] The application also provides a mobile air purifier path planning and obstacle avoidance system, comprising a multi-sensor group, an adjustable height sensor support, a mobile platform, a data fusion module, a path planning module, and a control unit.

[0036] The multi-sensor group comprises a two-dimensional laser radar, a ToF sensor, an RGB-D camera, and an air quality sensor; the two-dimensional laser radar, ToF sensor, and RGB-D camera are installed on the sensor support; the mobile platform is provided with a wheeled chassis.

[0037] The two-dimensional laser radar is used to obtain horizontal plane obstacle distance and angle data;

[0038] The ToF sensor and RGB-D camera are respectively used to collect obstacle height and three-dimensional profile data in real time;

[0039] The air quality sensor is used to collect PM2.5 and formaldehyde pollution concentration data of the current environment;

[0040] The data fusion module is used to fuse the data of the two-dimensional laser radar and the ToF sensor by using an extended Kalman filter algorithm, generate a three-dimensional voxel map, and the voxel resolution is 10cm×10cm×10cm; each voxel contains an occupancy state and a height value;

[0041] The path planning module is used for:

[0042] Passing judgment:

[0043] According to the safety margin set by the size of the air purifier, the area with a height <H+h or a horizontal size <L+l / W+w is marked as an impassable area; wherein, L, W, H are the length, width, and height of the air purifier respectively; h is the height safety margin, and l, w are the horizontal safety margins corresponding to the length and width of the air purifier respectively;

[0044] Global planning:

[0045] The A* algorithm is used for global path planning:

[0046] The heuristic function is dynamically adjusted according to the complexity of the environment; in a simple environment, the Manhattan distance is used to accelerate the search; in a complex environment, the Euclidean distance and dynamic weight combination are switched;

[0047] When planning the path, the impassable area is avoided, and the path through the pollution hot area is preferentially selected;

[0048] Local obstacle avoidance:

[0049] Velocity sample generation: based on the current velocity, 50 groups of velocity samples are generated within the range of ±20% of the current velocity;

[0050] Trajectory simulation: for each group of velocity samples, the trajectory of the future 2s is simulated;

[0051] High filtering: remove samples in the trajectory that pass through the area with a height <H+h;

[0052] Weighted scoring: the remaining samples are scored according to safety 40%, smoothness 30%, and target approach 30%;

[0053] Select the optimal trajectory: according to the weighted scoring result, select the optimal trajectory as the current local obstacle avoidance path;

[0054] Task linkage:

[0055] Based on the air quality sensor data, the DBSCAN clustering algorithm is used to identify areas with a concentration greater than a threshold, and a hot area priority queue is generated; the threshold is PM2.5>75μg / m 3 or formaldehyde>0.08mg / m 3 ;

[0056] Determine whether a pollution hot area is detected:

[0057] If yes: interrupt the current path, insert a hot area purification task, and use the dynamic window method to preferentially plan the shortest collision-free path to the hot area;

[0058] If not: continue to execute the current path planning;

[0059] The control unit is used for:

[0060] The path planning result is converted into a motor control signal of the wheeled chassis through a PID control algorithm, so that the air purifier moves according to the planned path.

[0061] During the movement of the air purifier, real-time path planning and obstacle avoidance adjustment are realized according to the data collected by the sensor in real time, so that a closed-loop control process is formed.

[0062] The application further provides a mobile air purifier comprising the mobile air purifier path planning and obstacle avoidance system.

[0063] Through the above scheme, the mobile air purifier path planning and obstacle avoidance method, system and mobile air purifier adopt a three-dimensional path planning scheme based on laser radar and multi-sensor fusion, construct a dynamic passable area model, combine a hierarchical obstacle avoidance strategy and task priority scheduling, and can realize safe and efficient autonomous navigation and precise purification.

[0064] The above description is only a summary of the technical scheme of the application, in order to more clearly understand the technical means of the application, and the content of the specification can be implemented, the following preferred embodiments of the application are described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0065] Figure 1 A flowchart of the mobile air purifier path planning and obstacle avoidance method in an embodiment of the application is shown in the figure.

[0066] Figure 2 A three-dimensional voxel map schematic diagram (division logic of visual passable area and impassable area) in an embodiment of the application is shown in the figure.

[0067] Figure 3 A local obstacle avoidance trajectory simulation diagram in an embodiment of the application is shown in the figure.

[0068] Figure 4 A structural schematic diagram of the mobile air purifier path planning and obstacle avoidance system of the application is shown in the figure. DETAILED DESCRIPTION

[0069] The specific embodiments of the application will be further described in detail below with reference to the accompanying drawings and embodiments. The following embodiments are used to illustrate the application, but not to limit the scope of the application.

[0070] Term explanation:

[0071] Three-dimensional voxel map: a three-dimensional environmental model constructed through multi-sensor data fusion, each voxel unit records the spatial occupancy state and height information, used to accurately describe the spatial distribution of obstacles.

[0072] Dynamic window approach (DWA): a local obstacle avoidance algorithm, generates feasible speed samples and simulates trajectories, combined with a high-pass filtering mechanism to filter safe paths, ensuring the smoothness and safety of real-time obstacle avoidance.

[0073] A* algorithm: a global path planning algorithm, dynamically adjusts the search strategy through heuristic function, calculates the optimal path in the three-dimensional passable area, balances the search efficiency and path quality.

[0074] Safety margin: a buffer parameter (h, l, w) based on the physical dimensions of the device (length L, width W, height H), used to compensate for shaking and turning requirements during movement, defines the minimum safe passage space (H+h, L+l, W+w).

[0075] Air quality hot zone: a high pollution concentration area identified through cluster analysis, linked to the path planning algorithm to prioritize purification tasks and improve regional purification efficiency.

[0076] Participate Figures 1 to 3 As shown in the embodiment, a laser radar and height perception-based mobile air purifier path planning and obstacle avoidance method is provided, including the following steps:

[0077] Step 1, multi-sensor group data acquisition:

[0078] Obtain horizontal plane obstacle distance and angle data through two-dimensional laser radar, real-time collect obstacle height and three-dimensional contour data through ToF sensor and RGB-D camera, collect current environmental PM2.5 and formaldehyde pollution concentration data through air quality sensor.

[0079] Step 2, space-time calibration:

[0080] Calibrate the data collected in step 1 in space-time to ensure consistency of data collected by different sensors in time and space, control error within 5ms.

[0081] Step 3, fusion modeling:

[0082] Use extended Kalman filter algorithm to fuse data from two-dimensional laser radar and ToF sensor, generate three-dimensional voxel map, voxel resolution is 10cm×10cm×10cm, each voxel contains occupancy state (0 / 1) and height value (accuracy ±2cm).

[0083] Step 4, passage judgment:

[0084] According to the safety margin (h=5cm, l=3cm, w=3cm) set according to the size of the air purifier, the area with a marked height <H+h or a horizontal size <L+l / W+w is an impassable area; wherein L, W, H are the length, width and height of the air purifier respectively; h is the height safety margin, and l, w are the horizontal safety margins corresponding to the length and width of the air purifier respectively. The minimum passable space is defined as (H+h, L+l, W+w), wherein h is the height safety margin (5-10cm), and l, w are the horizontal safety margins (3-5cm).

[0085] Step 5, global planning:

[0086] The A* algorithm is used for global path planning:

[0087] The heuristic function is dynamically adjusted according to the complexity of the environment. In a simple environment, the Manhattan distance is used to accelerate the search; in a complex environment, the Euclidean distance and dynamic weight combination are switched. The weight factors are α=0.6 and β=0.4, and the path optimization rate is improved by 18%;

[0088] When planning the path, avoid the impassable area and preferentially select the path through the pollution hot area.

[0089] Step 6, local obstacle avoidance (DWA+height filtering):

[0090] Velocity sample generation: based on the current velocity, 50 groups of velocity samples are generated within the range of the current velocity ±20%;

[0091] Trajectory simulation: simulate the trajectory of the future 2s for each group of velocity samples;

[0092] Height filtering: eliminate samples in the trajectory that pass through the height <H+h area;

[0093] Weighted scoring: the remaining samples are scored according to safety 40%, smoothness 30%, and target approach 30%;

[0094] Select the optimal trajectory: according to the weighted score result, select the optimal trajectory as the current local obstacle avoidance path (real-time adjust the speed and direction to avoid dynamic obstacles). The trajectory of the present application only retains the path with a height ≥45cm, and the traditional DWA trajectory contains a path with insufficient height.

[0095] Step 7, task linkage:

[0096] Pollution hot area identification: based on air quality sensor data, the DBSCAN clustering algorithm is used to identify areas with a concentration greater than a threshold to generate a hot area priority queue; the threshold is PM2.5>75μg / m 3 or formaldehyde >0.08mg / m 3 .

[0097] Path re-planning:

[0098] Judge whether the contaminated hot zone is detected:

[0099] If yes: interrupt the current path, insert the hot zone purification task, and use the dynamic window method to plan the shortest collision-free path to the hot zone;

[0100] If no: continue to execute the current path planning;

[0101] Step 8, execute control:

[0102] Convert the path planning result into motor control signals of the wheeled chassis through the PID control algorithm, and drive the air purifier to move according to the planned path;

[0103] Step 9, loop execution:

[0104] The air purifier continuously returns to step 1 for data acquisition during movement, constantly updates environmental information, realizes real-time path planning and obstacle avoidance adjustment, and forms a closed-loop control process.

[0105] Referring to Figure 4 The embodiment also provides a mobile air purifier path planning and obstacle avoidance system, which comprises a multi-sensor group, an adjustable height sensor support, a mobile platform, a data fusion module, a path planning module and a control unit.

[0106] The multi-sensor group comprises a two-dimensional laser radar, a ToF sensor, an RGB-D camera and an air quality sensor; the two-dimensional laser radar, the ToF sensor and the RGB-D camera are installed on the sensor support.

[0107] The mobile platform is provided with a wheeled chassis, and the wheeled chassis adopts a differential drive structure and has a minimum turning radius of 30 cm, which is suitable for narrow spaces.

[0108] The sensor support is adjustable in height (adjustment range 30-50 cm), so that the laser radar and the height sensor can cover the height (H=40 cm) of the equipment main body.

[0109] The two-dimensional laser radar is a 16-line laser radar (precision ±2 cm, scanning range 360°x10 m), which is used to obtain horizontal plane obstacle distance and angle data.

[0110] The height perception module: the ToF sensor (precision ±1 cm, range 0.1-5 m) and the RGB-D camera (depth resolution 1 mm) are respectively used to collect obstacle height and three-dimensional contour data in real time.

[0111] The air quality sensor is a PM2.5 / formaldehyde composite sensor (response time <10s), which is used to collect the pollution concentration data of PM2.5 and formaldehyde in the current environment, so as to locate the pollution hot area.

[0112] The data fusion module is used to fuse the data of the two-dimensional laser radar and the ToF sensor by using an extended Kalman filtering algorithm, to generate a three-dimensional voxel map, the voxel resolution being 10cmx10cmx10cm, and each voxel containing an occupancy state and a height value.

[0113] The path planning module is used to perform the above steps 4 to 7.

[0114] The control unit is used to perform the above steps 8 and 9.

[0115] Embodiment:

[0116] 1. System parameters (see Table 1)

[0117] Table 1 System parameters

[0118]

[0119]

[0120] 2. Test data (see Table 2)

[0121] Table 2 Test data

[0122]

[0123] Laboratory verification:

[0124] 1. Three-dimensional mapping accuracy test

[0125] Test environment: a 100m2 room containing a table and chair (height 70cm) and a sofa (bottom height 15cm).

[0126] Results: The voxel map height error is ≤3cm, and the sofa bottom non-passable area can be accurately identified (the traditional scheme misjudges the passable rate by 18%).

[0127] 2. Obstacle avoidance reliability test

[0128] Obstacle type: columnar obstacle (height 30cm, diameter 20cm), suspended obstacle (height 50cm). Results: The columnar obstacle avoidance success rate is 100%, and the suspended obstacle misjudgment rate is 0 (the traditional scheme misjudgment rate is 35%).

[0129] 3. Task linkage efficiency test

[0130] Pollution hot area setting: PM2.5 concentration 150ug / m3, formaldehyde concentration 0.1mg / m3.3 Area (area 10 m2).

[0131] Results: The priority planning path takes 2.3s, which is 30% less than the traditional scheme (no hot zone scheduling), and the hot zone purification time is shortened by 40%.

[0132] The application adopts a three-dimensional path planning scheme based on laser radar and multi-sensor fusion, which has the following technical effects:

[0133] 1) By constructing a dynamic passable area model, combining hierarchical obstacle avoidance strategy and task priority scheduling, safe and efficient autonomous navigation and precise purification can be realized.

[0134] 2) The three-dimensional voxel map dynamically defines the passable area in combination with the device size, breaking through the limitations of traditional two-dimensional obstacle avoidance.

[0135] 3) A* algorithm and DWA fusion high filtering mechanism, build "global planning-local fine adjustment-task linkage" closed loop.

[0136] 4) Modular sensor configuration and engineering safety margin design, adapt to various home environments. Prototype test shows that the complex scene pass success rate is 97%, and the purification efficiency is improved by 22%.

[0137] Compared with the prior art, as shown in Table 3.

[0138] Table 3 Comparison table with prior art

[0139]

[0140]

[0141] The embodiment also provides a mobile air purifier comprising the mobile air purifier path planning and obstacle avoidance system.

[0142] The above is only the preferred embodiment of the present application, and is not used to limit the present application, it should be pointed out that, for ordinary skilled in the art, without departing from the technical principles of the present application, can make several improvements and modifications, these improvements and modifications should also be considered as the protection scope of the present application.

Claims

1. A mobile air purifier path planning and obstacle avoidance method, characterized in that, Comprise the following steps: Step 1, multi-sensor group data acquisition: Obtain the horizontal plane obstacle distance and angle data through the two-dimensional laser radar, collect the obstacle height and three-dimensional profile data in real time through the ToF sensor and RGB-D camera, and collect the current environmental PM2.5 and formaldehyde pollution concentration data through the air quality sensor; Step 2, space-time calibration: The data collected in step 1 is calibrated in time and space to ensure that the data collected by different sensors is consistent in time and space, and the error is controlled within 5ms; Step 3, fusion modeling: The extended Kalman filter algorithm is used to fuse the data of two-dimensional laser radar and ToF sensor to generate a three-dimensional voxel map, and each voxel contains an occupancy state and a height value with a resolution of 10cm*10cm*10cm; Step 4, passage judgment: According to the safety margin set by the size of the air purifier, mark the area with a height <H+h or a horizontal size <L+l / W+w as an impassable area; Wherein, L, W, H are the length, width and height of the air purifier respectively; h is the height safety margin, and l, w are the horizontal safety margins corresponding to the length and width of the air purifier respectively; Step 5, global planning: Global path planning is performed using A* algorithm: According to the complexity of the environment, the heuristic function is dynamically adjusted, Manhattan distance is used to accelerate the search in simple environment, and Euclidean distance and dynamic weight combination are used in complex environment; Avoid impassable areas when planning the path, and preferentially select the path through the pollution hot area; Step 6, local obstacle avoidance: Velocity sample generation: Based on the current velocity, 50 groups of velocity samples are generated within the range of current velocity ±20%; Trajectory simulation: Simulate the trajectory of each group of velocity samples for 2s in the future; Height filtering: Remove samples in the trajectory that pass through the area with height <H+h; Weighted scoring: The remaining samples are scored according to safety 40%, smoothness 30% and target approach 30%; Select the optimal trajectory: According to the weighted score result, select the optimal trajectory as the current local obstacle avoidance path; Step 7, task linkage: Based on the air quality sensor data, the DBSCAN clustering algorithm is used to identify the area with concentration greater than the threshold value, and a hot area priority queue is generated; the threshold value is PM2.5>75μg / m 3 Or formaldehyde>0.08mg / m 3 ; Determine whether the pollution hot area is detected: If yes: Interrupt the current path, insert the hot area purification task, and use the dynamic window method to preferentially plan the shortest collision-free path to the hot area; If not: Continue to execute the current path planning; Step 8, execution control: Convert the path planning result into motor control signal of the wheeled chassis through PID control algorithm to drive the air purifier to move according to the planned path; Step 9, loop execution: The air purifier continuously returns to step 1 for data acquisition during movement, constantly updates environmental information, realizes real-time path planning and obstacle avoidance adjustment, and forms a closed-loop control process.

2. A mobile air purifier path planning and obstacle avoidance system, characterized in that, Comprise a multi-sensor group, an adjustable height sensor support, a mobile platform, a data fusion module, a path planning module, and a control unit; The multi-sensor group comprises a two-dimensional laser radar, a ToF sensor, an RGB-D camera, and an air quality sensor; The two-dimensional laser radar, ToF sensor, and RGB-D camera are installed on the sensor support; The mobile platform is provided with a wheeled chassis; The two-dimensional laser radar is used for acquiring horizontal plane obstacle distance and angle data; The ToF sensor and the RGB-D camera are respectively used for collecting obstacle height and three-dimensional profile data in real time; The air quality sensor is used for collecting PM2.5 and formaldehyde pollution concentration data of the current environment; The data fusion module is used for fusing data of the two-dimensional laser radar and the ToF sensor by using an extended Kalman filtering algorithm, generating a three-dimensional voxel map, and each voxel includes an occupancy state and a height value, and a voxel resolution is 10 cm*10 cm*10 cm; The path planning module is used for: Passage judgment: According to a safety margin set according to the size of the air purifier, an area with a height <H+h or a horizontal size <L+l / W+w is marked as an impassable area; wherein L, W, H are respectively the length, width and height of the air purifier; h is a height safety margin, and l, w are respectively horizontal safety margins corresponding to the length and width of the air purifier; Global planning: An A* algorithm is used for global path planning: A heuristic function is dynamically adjusted according to the complexity of the environment, and Manhattan distance is used to accelerate the search in a simple environment; and Euclidean distance and dynamic weight combination are switched in a complex environment; An impassable area is avoided when planning a path, and a path passing through a pollution hot area is preferentially selected; Local obstacle avoidance: Velocity sample generation: 50 groups of velocity samples are generated within the range of the current velocity ±20% based on the current velocity; Trajectory simulation: the trajectory of each group of velocity samples is simulated for 2s in the future; Height filtering: samples passing through the area with a height <H+h in the trajectory are removed; Weighted scoring: the remaining samples are scored according to safety 40%, smoothness 30% and target approach 30% by weight; Selecting the optimal trajectory: according to the weighted score result, the optimal trajectory is selected as the current local obstacle avoidance path; Task linkage: Based on the air quality sensor data, the DBSCAN clustering algorithm is used to identify the area with concentration greater than the threshold value, and a hot area priority queue is generated; the threshold value is PM2.5>75μg / m 3 Or formaldehyde>0.08mg / m 3 ; Judging whether a pollution hot area is detected: If yes: interrupting the current path, inserting a hot area purification task, and preferentially planning a shortest collision-free path to the hot area by using a dynamic window method; If no: continuing to execute the current path planning; The control unit is used for: Converting the path planning result into a motor control signal of the wheeled chassis by using a PID control algorithm, and driving the air purifier to move according to the planned path; According to the data collected in real time by the sensor during the movement of the air purifier, real-time path planning and obstacle avoidance adjustment are realized, and a closed-loop control process is formed.

3. A mobile air purifier, characterized by, The mobile air purifier path planning and obstacle avoidance system of claim 2 is included.

Images