A method for remotely manually relocating coordinates of an indoor robot

Abstract

The application relates to the technical field of robots and discloses a method for remotely manually repositioning coordinates of an indoor robot, which comprises the following steps: step 1, identifying a positioning loss state, and the platform sends an action: when the robot cannot maintain reliable positioning, the platform timely sends an alarm, and after an operator receives and confirms the alarm, the repositioning process is started; step 2, constructing a new map: a new map construction mechanism is set; step 3, calculating a target pose: the operator manually aligns the new map and the old map to obtain the target pose; and step 4, repositioning: the robot realizes internal updating according to the target pose. The application combines intelligent guidance of an automatic algorithm and flexible decision-making of manual operation, effectively overcomes the defects that pure automatic repositioning is prone to failure in a large range loss or a complex environment and falls into a local optimum, and simultaneously avoids the problems that traditional manual intervention needs on-site operation and is low in efficiency.

Description

Technical Field

[0001] This invention relates to the field of robotics, and more particularly to a method for remotely and manually repositioning the coordinates of an indoor robot. Background Technology

[0002] Mobile robots are robotic systems that can move and operate autonomously or semi-autonomously in their environment. They are typically equipped with sensors, control systems, drive systems, and communication modules, and can perform various tasks without direct human intervention.

[0003] In mobile robot localization systems based on laser SLAM technology, although it has demonstrated high efficiency in indoor navigation, it still faces many challenges in complex and ever-changing real-world environments. For example, in dynamic interference environments, such as frequently moving obstacles obscuring the laser feature points originally used for localization, or in areas with sparse features, such as long corridors, the lack of sufficient environmental information for self-localization updates can lead to decreased localization accuracy or even loss of localization. In addition, sudden hardware failures, such as packet loss during radar data transmission, can also severely affect the robot's normal localization function. Once these situations occur, the robot may be unable to accurately determine its own position, thus triggering the so-called "kidnapping detection" state. At this time, the robot loses accurate and reliable pose estimation on the map. Therefore, it is necessary to study relocalization techniques for robots when facing the above problems. Summary of the Invention

[0004] The purpose of this invention is to enable remote manual repositioning of the robot by obtaining its map information remotely via the network after the robot loses its localization, and then manually matching the map.

[0005] To achieve the above objectives, the present invention adopts the following technical solution:

[0006] A method for remote manual repositioning of an indoor robot includes the following steps: Step 1, identifying the location loss status and reporting it to the platform; Step 2, after receiving and confirming the alarm, the operator initiates the repositioning process; Step 3, the robot begins building a new map; Step 4, the platform recommends the best scanning area; Step 5, the operator refines the new map data based on the recommended mobile robot; Step 6, the robot reports the old and new map data; Step 7, the platform visualizes the old and new maps; Step 8, the operator manually aligns the old and new maps; Step 9, the target pose is calculated; Step 10, the platform issues a pose correction command; Step 11, the robot forcibly updates its pose and completes the repositioning.

[0007] As a preferred technical solution of the present invention, in step 1, the positioning loss state includes a sharp drop in positioning confidence, unreasonable jumps in pose estimation, and SLAM algorithm reporting failure; when the loss state occurs, the robot immediately stops autonomous movement, maintains lidar scanning and communication functions, and sends a positioning loss event message to the central management platform.

[0008] As a preferred technical solution of the present invention, step 3 specifically includes the following steps: Reset: clear or pause data related to the old map; Initialize new map: start a new SLAM instance with the robot's current physical position as the origin of the new coordinate system; Generate new data: start building a new map, and the data of the LiDAR and encoder disk in the robot continue to be collected, record its pose in the new map coordinate system in real time, and transmit it to the platform.

[0009] As a preferred technical solution of the present invention, step 4 includes data acquisition: collecting old map data; feature extraction: using a combination of line segment extraction and intersection point extraction to perform image processing, performing edge detection on the old map, extracting line segments from the edges, and calculating the line segment intersection points; region filtering and sorting: clustering the extracted high feature points to form several high feature regions; visual recommendation: marking on the platform map interface, and forming recommended paths according to the robot's last reported pose.

[0010] As a preferred embodiment of the present invention, in step 5, if the robot fails to reach the recommended high-feature area, a progressive spiral search mechanism is set to expand the search range.

[0011] As a preferred technical solution of the present invention, in step 5, the progressive spiral search mechanism includes: setting the spiral density and angle increment, wherein the spiral density is less than the robot's lidar recognition distance, taking the current physical position as the starting point, generating an Archimedean spiral, and the operator controls the robot to move and scan step by step along the route until it enters one of the high feature areas.

[0012] As a preferred technical solution of the present invention, in step 7, the old map and the new map are loaded simultaneously in the map display engine, and the two maps are rendered with different colors and line types to distinguish them.

[0013] As a preferred technical solution of the present invention, in step 8, the position of the old map layer remains unchanged, the operator uses a tool to drag the new map layer to translate it, and uses a rotation control to rotate the new map layer until the corresponding parts in the map completely overlap.

[0014] As a preferred technical solution of the present invention, in step 8: when the new and old maps only partially overlap and the rest are severely misaligned or completely mismatched, the operator instructs the robot to start from the currently aligned position and move towards the misaligned direction to perform a small-range supplementary scan, and then return to step 6; when there are multiple solutions for aligning the new and old maps, the system directly returns to step 4 to increase the threshold of the feature extraction algorithm.

[0015] As a preferred technical solution of the present invention, in step 9, the platform system records the total geometric transformation performed by the operator on the new map in real time, records the translation vector and rotation angle of the new map, and obtains the target pose.

[0016] The beneficial effects of this invention are as follows:

[0017] The indoor robot remote manual relocalization method proposed in this invention cleverly integrates the intelligent guidance of automatic algorithms with the flexible decision-making of manual operation compared to existing technologies. It effectively overcomes the shortcomings of purely automatic relocalization, such as easy failure and getting trapped in local optima in large-scale loss or complex environments. At the same time, it avoids the problems of traditional manual intervention requiring on-site operation and low efficiency. By intelligently recommending the best scanning area through the platform and providing visual navigation, it significantly reduces the judgment difficulty and operation threshold for operators. It introduces a progressive spiral search as a fallback strategy to ensure that the environment exploration can still be completed systematically when the recommended path fails. It also designs a closed-loop processing mechanism based on alignment result feedback, which can improve the alignment success rate by supplementing scanning or increasing feature thresholds. The whole solution achieves efficient remote intervention while ensuring high reliability and strong robustness, significantly improving the autonomy and operation and maintenance efficiency of the robot system. Attached Figure Description

[0018] Figure 1 This is a schematic diagram of a method for remote manual repositioning of coordinates of an indoor robot proposed in this invention.

[0019] Figure 2 This is a flowchart of a method for remote manual repositioning of coordinates of an indoor robot proposed in this invention;

[0020] Figure 3 A flowchart for setting the recommended scan path;

[0021] Figure 4 A system diagram of the progressive spiral search mechanism;

[0022] Figure 5 The display of new and old maps uploaded to the robot on the platform;

[0023] Figure 6 This is a display image showing the new map aligned with the old map.

[0024] Figure 7 This is a physical image of the robot of the present invention. Detailed Implementation

[0025] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0026] Before introducing the technical solutions described in this specification, a brief introduction to existing indoor robot repositioning methods will be given;

[0027] When a robot triggers a "kidnapping detection" state, traditional solutions generally fall into two categories:

[0028] 1. Automatic Relocalization Algorithms: These methods aim to allow robot systems to recover from localization loss without human intervention. The core idea is to re-estimate the robot's pose by matching current sensor data with a known map. These generally include:

[0029] The probabilistic relocalization method based on particle filtering: The environment is modeled as a series of possible positions, called particles. Each particle is assigned a weight according to the degree of matching between the current laser scan data and the map. Through iterative updates, particles with high weights gradually converge to the true pose. It is suitable for relocalization of completely unknown initial positions. However, in symmetrical or repetitive structural environments, such as multiple identical rooms or long corridors, the matching scores of multiple positions are similar, leading to "particle degradation", that is, a large number of particle weights converge and cannot converge effectively.

[0030] Scan matching algorithms such as ICP and its variants: By iterating the nearest point or its improved version such as NDT, the current laser scan data is registered with the scan data in the map, and the optimal transformation matrix is ​​solved to recover the pose. This method has high accuracy when the features are rich and the initial pose is close to the true value. However, this method is highly sensitive to the initial pose. If the robot completely loses its localization, that is, the initial guessed pose is greatly deviated, it is easy to get stuck in local optima and the matching will fail. Especially in large-scale maps, the matching process takes a long time and performs poorly in low feature areas.

[0031] Multi-sensor fusion solutions combine information from multiple sensors such as LiDAR, IMU (Inertial Measurement Unit), wheel encoders, and vision cameras to improve the redundancy and robustness of the positioning system. When a single sensor fails, such as when the radar is blocked, other sensors can provide auxiliary information, which can effectively suppress noise and improve stability in dynamic environments. However, the hardware cost increases significantly, requiring the additional purchase and integration of multiple sensors. At the same time, the system complexity is high, requiring the design of complex fusion algorithms such as EKF, UKF, and graph optimization to coordinate the data from different sensors.

[0032] 2. Manual intervention solution: The operator goes to the location of the robot and triggers the reset or repositioning mode through a physical button to achieve manual repositioning on site, but the timeliness of the operation is poor.

[0033] In summary, for automatic relocation algorithms, the program automatically calculates the relocation, which requires calculation within a limited search space. If the search range is incorrect, it cannot be located. If it is located at a similar but incorrect location, it will take too long to calculate and will time out and fail, meaning there is a possibility of location failure. On the other hand, manual intervention requires personnel to go to the site, which is time-consuming and labor-intensive, and frequent intervention will affect operational efficiency.

[0034] In addition, before introducing the solution of this invention, a brief introduction to the robot system is given. The robot system consists of an end robot and a central management platform. The behavior of the end robot includes: the LiDAR continuously scans the surrounding environment to obtain obstacle point cloud data; the encoder or encoder on the action wheel records the robot's wheel speed and mileage information in real time; the SLAM (Simultaneous Localization and Mapping) algorithm inside the robot fuses the LiDAR data and encoder data to perform real-time self-localization and determine its own pose on the constructed global map, specifically the position coordinates x, y and the orientation angle θ. The platform behavior is that the platform continuously receives and displays the robot's real-time position, status and map information.

[0035] Based on the above description, please refer to the appendix. Figures 1-7 This invention proposes a method for remote manual repositioning of an indoor robot, comprising the following steps:

[0036] Step 1: Identify and report the location loss status to the platform: When the robot cannot maintain reliable positioning, it promptly sends an alarm to the platform and initiates the repositioning process. Specifically, when the robot cannot maintain reliable positioning due to dynamic obstacle occlusion, sparse feature areas, or temporary sensor failure, it is determined to be a location loss. The robot's internal positioning monitoring module detects a sharp drop in positioning confidence, unreasonable jumps in pose estimation, or SLAM algorithm failure. At this time, it is confirmed to enter the "kidnapping detection" state. To avoid collisions, the robot immediately stops autonomous movement but maintains LiDAR scanning and communication functions. It sends a location loss event message to the central management platform through communication modules such as Wi-Fi, 4G, and 5G. The message includes the robot ID, timestamp, and the last reported low-confidence pose.

[0037] Step 2, Platform Actions: After the operator receives and confirms the alarm, the relocation process is officially initiated, instructing the robot to enter a specific mode. Specifically, after the platform receives the location loss information, it prominently prompts the operator on the interface that a certain robot has lost its location. For example, the robot is highlighted in a color on the user interface, and a notification pops up requesting manual intervention. After the operator confirms the alarm, the relocation process is initiated. The platform generates and sends a control message containing the "enter_manual_relocalization_mode" instruction to the target robot. The robot receives and parses the instruction and confirms that it has entered the manual relocation mode.

[0038] Step 3: The robot begins building a new map: It re-establishes an independent local coordinate system at the current location to provide a basis for subsequent alignment. This includes:

[0039] Reset: After receiving the instruction, the robot clears or pauses the SLAM calculation state and map data related to the old map;

[0040] Initialize a new map: Start a new SLAM instance with the robot's current physical position as the origin (0, 0, 0) of the new coordinate system. This instance can use the same algorithm as the main SLAM, such as the Cartographer algorithm.

[0041] Generate new data: Start building an independent local map, called the new map. LiDAR and encoder data continue to be collected, and the pose (x_new, y_new, θ_new) in the new map coordinate system is recorded in real time and transmitted to the platform.

[0042] Step 4: Improve the new map by calculating and recommending the best scanning area based on the platform: Utilize algorithms to intelligently analyze the old map and combine it with the robot's position to identify the most advantageous scanning target area for rapid and accurate relocation, and guide the operation. Please refer to the appendix for details. Figure 3 ,include:

[0043] Data Acquisition: The platform already has an old map, which can be used in Occupancy Grid Map format, as well as the approximate location (x_last, y_last) last reported by the robot;

[0044] Feature extraction: A combination of line segment extraction and intersection or endpoint extraction is used. The OpenCV library or PCL (PointCloud Library) is used for image or grid processing. Edge detection or direct contour extraction is performed on the binarized raster image of the old map. Line segment detection algorithms such as Probabilistic Hough Transform are used to extract straight line segments from the edges. The intersection and endpoints of all line segments are calculated. These points usually correspond to strong features such as corners, door frames, and pillars.

[0045] Optionally, the curvature or corner response of local regions, such as in a variant of the Harris Corner Detector, can be calculated to supplement the identification of non-linear features;

[0046] Region filtering and sorting: The extracted high-feature points are clustered to form several high-feature regions. Specifically:

[0047] Set the current search range, for example, 10m long and 5m wide; set the increment, for example, the length increases by 2m and the width increases by 1m each time the search is performed; set the feature threshold, for example, the number of feature points in the region is greater than 2.

[0048] Using (x_last, y_last) as the search center, the search is performed according to the set current search range. High feature regions within the range are filtered out. The filtered regions are weighted and scored according to feature intensity (feature point density within the region) and distance (distance to (x_last, y_last)). Regions that are close and have strong features have the highest priority. If there are no high feature regions within the current search range, the range is expanded according to the set increment.

[0049] Visualized Recommendations: On the platform map interface, the optimal recommended area is marked with a prominent graphic, and a navigation arrow points from the robot's current position to the recommended area.

[0050] Step 5: Operator remotely controls the robot based on recommendations: The operator uses the platform's intelligent recommendations to direct the robot to the optimal scanning position, receives and parses remote movement commands from the platform, executes the commands, and drives the movement wheels to move towards the target area. In addition, during the movement, the LiDAR continues to scan, the encoder continuously records mileage, and the new SLAM instance continues to work, integrating newly collected data to continuously expand and update the new map, ensuring that it contains rich features of the recommended area. During the movement, the operator needs to closely observe the new map and sensor data to ensure that the robot moves in the correct direction and adjusts the direction and path according to the actual situation. For example, when encountering obstacles, the operator controls the robot to bypass the obstacles, or when the operator identifies obvious and sufficiently rich features based on the real-time uploaded map data, the search is paused and the next step is initiated.

[0051] It is worth noting that if the robot fails to reach the recommended high-feature area, for example, by encountering an obstacle or discovering that the area has been altered, the operator should expand the search area.

[0052] For further details, please refer to the appendix. Figure 4An Archimedean spiral search mechanism is set up, with the spiral density b and angle increment Δθ set. If the robot's LiDAR recognition distance is r, then b is less than r, ensuring sufficient overlap between adjacent scanning circles to avoid missing feature areas. The current position (x_current, y_current) is used as the starting point, and an Archimedean spiral is generated using the formula r = bθ. The robot is then controlled to move and scan along this route step by step until it enters one of the high-feature areas or the operator determines that the features in the map are sufficient.

[0053] Step 6: Robot reports new and old map data: Upload the key data used for alignment to the platform to provide materials for subsequent manual alignment. Specifically, when the operator believes that the new map is sufficient or the robot has reached the recommended area and completed the scan, the robot uploads the map data to the platform. The robot uploads the old map (the global map file before the positioning was lost) and the current complete new map (the local map file built in manual repositioning mode) to the server via the network. Optionally, the robot pose in the current coordinate system of the new map is also uploaded as a reference. The platform server receives and stores the data.

[0054] Step 7: Platform Visualization of Old and New Maps: This provides operators with a clear visual comparison environment, facilitating map alignment. Specifically, the old and new maps are loaded simultaneously in the map display engine, and the two maps are rendered using different colors and line types to distinguish them. Please refer to the appendix for details. Figure 5 In the image, the boundaries of the old map are shown in white, and the boundaries of the new map are shown in black.

[0055] Step 8: Operator manually aligns the old and new maps: Through manual intervention, the two maps are precisely matched to determine their spatial transformation relationship. Specifically, the operator uses a mouse or touchscreen to drag and pan the new map layer, and uses the rotation controls to rotate the new map layer. Careful adjustments are made to ensure that geometric features such as walls, corners, and pillars in the new map completely overlap or are aligned to the maximum extent with their corresponding parts in the old map. Please refer to the appendix for details. Figure 6 The operator moves and rotates the black boundary until it coincides with the white boundary, at which point the new map is perfectly aligned with the old map.

[0056] It is worth noting that when the new and old maps only partially overlap, and the rest are severely misaligned or completely mismatched, for example, a corner, a section of corridor, or a pillar in the new map can perfectly overlap with the old map, but other parts of the new map cannot be found on the old map, or the direction and distance are completely wrong, the main reason is that the new map is not fully constructed or has drift: during the robot's movement, although the SLAM algorithm captures local features, due to the long movement distance, sparse environmental features, or slight slippage, the odometry accumulates errors, causing the overall structure of the new map to be distorted or translated. At this time, the operator commands the robot to start from the current (aligned) position and move to the misaligned direction to perform a small-range, high-precision supplementary scan, and then re-enter step 6 sequentially.

[0057] When there are multiple solutions or ambiguities in aligning the old and new maps, for example, the new map shows a T-junction while the old map shows three identical T-junctions, and it is impossible to determine which one is the only correct one. That is, there are a large number of areas with similar geometric structures in the environment, and the collected feature points are not unique in the global map. In this case, return directly to step 4 and increase the threshold of the feature extraction algorithm: for example, change feature points > 2 to feature points > 5, or require that it must contain more unique features such as non-right angles or special shapes.

[0058] Step 9: Calculate the offset and target pose: Based on the results of manual alignment, calculate the robot's precise position and orientation in the old map coordinate system. Specifically, the platform system records in real time the total geometric transformations performed by the operator on the new map.

[0059] Translation vector (Δx, Δy): The distance moved from the initial upload position on the new map to the final alignment position;

[0060] Rotation angle (Δθ): The amount of rotation from the initial direction to the final alignment direction, where angle normalization should be noted;

[0061] Calculate the target pose: Since the robot starts scanning from the origin (0, 0, 0) of the new map, its target pose (x_target, y_target, θ_target) in the old map coordinate system is equal to the recorded transformations:

[0062] x_target = Δx;

[0063] y_target = Δy;

[0064] θ_target = Δθ;

[0065] Δx, Δy, and Δθ represent the transformations required to change the new map from its initial upload location to a location aligned with the old map.

[0066] Step 10: The platform issues a pose correction command: The calculated precise pose is sent to the robot, forcing it to update its internal state. The calculated (x_target, y_target, θ_target) is packaged into a forced pose correction command, such as in JSON format: {"cmd": "force_pose", "x": x_target, "y": y_target, "theta": θ_target}).

[0067] Instructions are sent to the target robot via a secure communication channel such as TLS-encrypted MQTT. Upon receiving the instructions, the robot enters a standby state and prepares to update its pose.

[0068] Step 11: Force the robot to update its pose and complete relocalization: The robot receives the new global pose, reintegrates into the original map, and resumes normal functions. Specifically, after receiving the forced pose command, the robot immediately stops all current localization-related calculations, forcibly sets its internal pose estimate to (x_target, y_target, θ_target), notifies the SLAM system that the pose has been reset, the robot successfully relocalizes in the old map coordinate system, exits the manual relocalization mode, and restores normal SLAM tracking and autonomous navigation capabilities. The platform updates the robot's position on the interface, displays that the relocalization was successful, and can record operation logs.

[0069] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. A method for remote manual repositioning of coordinates of an indoor robot, characterized in that, Includes the following steps: Step 1: Identify location loss status and the platform issues an action: When the robot cannot maintain reliable positioning, it promptly sends an alarm to the platform. After the operator receives and confirms the alarm, the repositioning process is initiated. Step 2: Construct a new map: Set up a new map construction mechanism. Using the current physical location of the robot as the origin, and combining the robot's pose, construct a new map with high-feature regions according to the construction mechanism. Step 3: Calculate the target pose: The operator manually aligns the new map with the old map to obtain the target pose; Step 4: Perform relocalization: Based on the target pose, the robot performs internal updates to achieve the relocalization purpose; Setting the recommended scanning path in step 2 specifically includes: Data Acquisition: Collecting old map data; Feature extraction: A combination of line segment extraction and intersection point extraction is used for image processing. Edge detection is performed on the old map, line segments are extracted from the edges, and the intersection points of the line segments are calculated. Region filtering and sorting: The extracted high-feature points are clustered to form several high-feature regions; Visualized recommendation: Marked on the platform map interface, a recommended path is formed based on the robot's last reported pose.

2. The method for remote manual repositioning of coordinates of an indoor robot according to claim 1, characterized in that, In step 1, the localization loss state includes a sharp drop in localization confidence, unreasonable jumps in pose estimation, and SLAM algorithm reporting failure. When a loss occurs, the robot immediately stops moving autonomously, maintains its LiDAR scanning and communication functions, and sends a location loss event message to the central management platform.

3. The method for remote manual repositioning of coordinates of an indoor robot according to claim 2, characterized in that, In step 2, the new map building mechanism includes: Preparation: Begin creating a new SLAM instance; Set recommended scanning path: Analyze the old map, divide the high feature area, and combine it with the robot's last reported pose to form a recommended path; Improved map: Operators can remotely control the robot based on the recommended scan path.

4. The method for remote manual repositioning of coordinates of an indoor robot according to claim 3, characterized in that, The construction preparation in step 2 specifically includes: Reset: Clears or pauses data related to the old map; Initialize new map: Start a new SLAM instance with the robot's current physical position as the origin of the new coordinate system; Generate new data: Start building a new map. The LiDAR and encoder data inside the robot continue to be collected, and its pose in the new map coordinate system is recorded in real time and transmitted to the platform.

5. A method for remote manual repositioning of coordinates of an indoor robot according to claim 4, characterized in that, In the improved new map in step 2, when the robot is being guided to walk along the recommended path, the operator can change the robot's direction and path according to the actual situation. If the robot fails to reach the recommended high-feature area, a progressive spiral search mechanism is set to expand the search range.

6. The method for remote manual repositioning of coordinates of an indoor robot according to claim 5, characterized in that, The progressive spiral search mechanism includes: setting the spiral density and angle increment, with the spiral density being less than the robot's lidar recognition distance; using the current physical position as the starting point, generating an Archimedean spiral; and having the operator control the robot to move and scan along this path step by step until entering one of the high-feature regions.

7. The method for remote manual repositioning of coordinates of an indoor robot according to claim 6, characterized in that, Step 3 includes: The robot reports new and old map data: It uploads key data for alignment to the platform, providing materials for subsequent manual alignment. The platform visualizes the old and new maps: In the map display engine, the old map and the new map are loaded at the same time, and the two maps are rendered with different colors and line types to distinguish them; Operator manually aligns the old and new maps: Keeping the position of the old map layer unchanged, the operator uses the tool to drag the new map layer to translate it, and uses the rotation control to rotate the new map layer until the corresponding parts in the map completely overlap. Calculate offset and target pose: Based on the results of manual alignment, calculate the robot's precise position and orientation in the old map coordinate system.

8. A method for remote manual repositioning of coordinates of an indoor robot according to claim 7, characterized in that, In step 3, the operator manually aligns the old and new maps: When the old and new maps only partially overlap, and the rest are severely misaligned or completely mismatched, the operator instructs the robot to start from the currently aligned position and move towards the misaligned direction to perform a small-scale supplementary scan, and then return to the step of the robot reporting the old and new map data.

9. A method for remote manual repositioning of coordinates of an indoor robot according to claim 8, characterized in that, In step 3, where the operator manually aligns the old and new maps: if there are multiple solutions to the alignment of the old and new maps, return directly to step 2 and set the high feature area again.

Images