Fixed-point interactive projection control method for wheeled chassis robot

Abstract

The invention discloses a fixed-point interactive projection control method for a wheeled chassis robot, and the method comprises the steps: controlling the wheeled chassis robot to move to a target position, and transmitting a first control instruction to an interactive projection system integrated in a robot body when the robot is in a moving state, so as to enable the interactive projection system to maintain a non-working state; after the robot reaches the target position, if it is judged that the robot enters the static state, a second control instruction is sent to the interactive projection system; in response to the second control instruction, starting a positioning sensor to scan the interaction area, and determining first position information of at least one interaction object; based on the first position information, determining a projection reference point of the projection picture, and generating interaction picture data projection; and starting the visual sensor, capturing the action of the interactive object based on the region of interest determined by the first position information, updating interactive picture data, and when an interaction ending condition is met, controlling the interactive projection system to stop working, and controlling the robot to recover the moving ability.

Description

Technical Field

[0001] This invention belongs to the field of fixed-point interactive projection technology, specifically relating to a fixed-point interactive projection control method for a wheeled chassis robot. Background Technology

[0002] With the rapid development of robotics technology, wheeled chassis robots have been widely used in various fields such as service, guidance, education, and entertainment due to their excellent mobility and environmental adaptability. Traditional wheeled chassis robots mainly undertake movement, transportation, or navigation tasks, and their human-computer interaction methods are mostly based on voice dialogue, touch screen operation, or simple indicator light feedback. Although these interaction methods are mature, they are relatively simple in form, lack immersion and environmental integration capabilities, and are difficult to provide rich, dynamic, and engaging interactive experiences in open spaces.

[0003] To enhance the interactive capabilities of robots, projection technology has been introduced into the field of human-computer interaction. Fixed interactive projection systems have been deployed in exhibitions, shopping malls, children's playgrounds, and other settings. These systems use cameras to capture user movements and project interactive images onto a fixed wall or floor, enabling haptic interaction. However, these systems are limited by fixed installation locations and cannot be flexibly moved according to user location or scene requirements, resulting in a limited interactive range and a lack of personalized and responsive user experience.

[0004] To address mobility issues, the industry has begun integrating projection devices into mobile robot platforms, forming the concept of mobile projection robots. Related technological solutions have mainly evolved along two directions: Adding projectors and cameras directly to mobile robots to attempt interactive projection during robot movement or brief pauses. While this approach achieves mobility for the projection system, it generally overlooks the fundamental impact of robot movement itself on projection quality. Because the robot's chassis continuously vibrates, changes posture, and drifts during movement, the projected image suffers from severe blurring, jitter, and geometric distortion. This drastically reduces the accuracy and stability of interactive recognition, resulting in a poor user experience and potentially causing visual misleading or safety hazards due to unstable projection.

[0005] Some solutions attempt to dynamically compensate for projection image jitter caused by robot movement using high-precision inertial sensors, gimbal stabilization, or real-time image processing algorithms. These solutions are technically complex and costly, and the compensation algorithms suffer from latency, making it difficult to completely eliminate image distortion under high-speed movement or complex road conditions, thus posing challenges to system reliability.

[0006] The integration of fixed interactive projection systems with mobile robot platforms in existing technologies mainly presents the following technical contradictions and problems: while giving robots mobility, it is difficult to guarantee the high definition and stability of the projected image; to achieve accurate interaction while moving, complex sensors and compensation algorithms need to be introduced, which increases system cost and failure rate; interactive projection while moving poses a safety risk of misoperation or misjudgment due to image distortion, and also damages the user's immersive interactive experience.

[0007] Therefore, there is currently a lack of a technical solution that enables wheeled chassis robots to flexibly reach any interactive point and provide a high-precision, stable, and safe interactive projection experience that is comparable to or even better than that of fixed systems. Summary of the Invention

[0008] This invention provides a fixed-point interactive projection control method for a wheeled chassis robot, which solves the problems of blurred and distorted projected images, low interactive recognition accuracy, poor system security, and poor user experience caused by robot vibration, posture changes, and position drift when performing interactive projection directly on a mobile robot platform.

[0009] The technical solution adopted in this invention is as follows: A fixed-point interactive projection control method for a wheeled chassis robot includes: Control the wheeled chassis robot to move to the target position. When the robot is in the moving state, send a first control command to the interactive projection system integrated into the robot body so that the interactive projection system remains in a non-working state. Once the robot reaches the target location, it is determined whether the robot has entered a stationary state. If it is determined that the robot has entered the stationary state, a second control command is sent to the interactive projection system. In response to the second control command, the interactive projection system activates the positioning sensor to scan the interactive area around the robot, and determines the first position information of at least one interactive object in the robot coordinate system based on the scan data. Based on the first position information, the projection reference point of the projected image on the target projection surface is determined, and the corresponding interactive image data is generated. The projector is then controlled to project the interactive image data onto the target projection surface. The visual sensor is activated to capture the movements of the interactive object based on the region of interest determined by the first location information. The interactive screen data is updated according to the captured movements. When the interaction ends, the interactive projection system is controlled to stop working, and the robot is controlled to resume its mobility.

[0010] The fixed-point interactive projection control method for wheeled chassis robots used in this invention also has the following additional technical features: To keep the interactive projection system in a non-operating state, specifically including: Send an instruction containing a moving status identifier to the control server of the interactive projection system; In response to the instruction, the control server executes at least one of the following: Disconnect or disable power supply to the projector; switch the positioning sensor and / or the vision sensor to a low-power standby mode.

[0011] Determining whether the robot has entered a stationary state includes: Monitor the robot's wheel speed and attitude angle data; If the wheel speed remains zero and the change in the attitude angle is less than a first threshold within a preset time window, then the robot is determined to have entered the stationary state.

[0012] After determining that the robot has entered the stationary state, the process further includes: Trigger the robot's braking device or telescopic support legs to enhance static stability.

[0013] Determining the first position information of at least one interactive object in the robot coordinate system based on scan data includes: The positioning sensor is a lidar; The point cloud data acquired by the lidar is subjected to ground filtering and clustering processing to identify point cloud clusters corresponding to the interactive objects; Calculate the two-dimensional coordinates of the bottom center point of the point cloud cluster in the robot coordinate system, and use it as the first position information.

[0014] Controlling the projector to project the interactive image data onto the target projection surface includes: According to the preset scene configuration, the target projection surface is selected as the ground or the wall. If the ground is selected, the ground point corresponding to the first location information is used as the projection reference point; If a wall is selected, the vertical projection point of the first position information on the wall is calculated based on the robot's current position and orientation, and used as the projection reference point. Based on the pre-calibrated projector installation parameters and coordinate system transformation relationship, the interactive screen data is geometrically corrected to compensate for the screen distortion caused by the tilted installation of the projector.

[0015] The region of interest determined based on the first location information refers to a preset range centered on the projection coordinates of the first location information on the imaging plane of the visual sensor.

[0016] The updating of the interactive screen data refers to the control server generating corresponding visual feedback content in real time based on the action type identified by the visual sensor, and integrating it into the interactive screen data to control the projector to update the projected screen.

[0017] The interaction termination condition includes any of the following: The preset time has elapsed since the interactive projection was initiated; The visual sensor recognizes the preset end gesture or command; The robot receives a new movement task instruction sent from the outside.

[0018] This invention further discloses a wheeled chassis robot, comprising: A mobile chassis is used to provide the robot with mobility. The control unit is used to control the movement of the mobile chassis; An interactive projection system is integrated into the robot body, including positioning sensors, vision sensors, a projector, and a control server; The control unit is configured to perform the method described.

[0019] Due to the adoption of the above technical solution, the beneficial effects achieved by this invention are as follows: 1. In this invention, the projection system remains in a non-working state while the robot is moving, only activating upon response to commands after the robot comes to a standstill. This effectively isolates the projection system from direct interference caused by chassis vibration, posture changes, and positional uncertainties during robot movement. This allows the projector to operate on a completely stable physical platform, thereby completely eliminating image jitter, blurring, and geometric distortion caused by moving the projection, ensuring the inherent stability and clarity of the projected image, and achieving high-quality human-computer interaction.

[0020] Furthermore, the positioning sensor is activated to scan and determine the initial location information. Even when stationary, the positioning sensor (such as LiDAR) can acquire precise coordinates of objects within the interactive area. This location information is independent of ambient lighting conditions and unaffected by projected content, offering higher reliability and accuracy compared to purely visual solutions. This initial location information forms the spatial anchor for all subsequent interactive actions.

[0021] This information is not only used to determine the projection reference point of the projected image, ensuring that the interactive interface can be accurately presented around or centered on the interactive object, thus improving the intuitiveness and user-friendliness of the interaction; more importantly, it is directly used to guide the region of interest of the visual sensor, concentrating computing resources on the most relevant image area. This collaborative perception mechanism, from coarse-grained positioning to fine-grained recognition, significantly improves the real-time performance, accuracy, and overall energy efficiency of motion capture, achieving precise and smooth haptic interaction.

[0022] This invention features clear behavior: disabling projection during the movement phase avoids safety risks caused by misleading visuals; resetting the control system and restoring mobility after interaction ensures the continuity and automation of robot task execution, significantly enhancing the system's practicality and reliability in complex, unstructured environments. While endowing a wheeled chassis robot with mobility, it successfully provides it with high-quality, high-precision, and safe fixed-point interactive projection capabilities comparable to fixed systems, overcoming the long-standing coordination problem between expressiveness and stability in mobile interactive robots. Attached Figure Description

[0023] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings: Figure 1 This is a flowchart illustrating the fixed-point interactive projection control method for a wheeled chassis robot according to one embodiment of the present invention. Figure 2 This is a structural diagram of the wheeled chassis robot according to one embodiment of the present invention.

[0024] in: 1. Mobile chassis; 2. Interactive projection system; 21. Positioning sensor; 22. Vision sensor; 23. Projector; 24. Control server; 3. Control unit. Detailed Implementation

[0025] To more clearly illustrate the overall concept of the present invention, a detailed description will be provided below with reference to the accompanying drawings and examples.

[0026] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0027] like Figure 1 As shown, a fixed-point interactive projection control method for a wheeled chassis robot includes: S100: Control the wheeled chassis robot to move to the target position. When the robot is in the moving state, send a first control command to the interactive projection system integrated into the robot body so that the interactive projection system remains in a non-working state. S200: When the robot reaches the target position, it is determined whether the robot has entered a stationary state. If it is determined that the robot has entered the stationary state, a second control command is sent to the interactive projection system. S300: In response to the second control command, the interactive projection system activates the positioning sensor to scan the interactive area around the robot, and determines the first position information of at least one interactive object in the robot coordinate system based on the scan data; S400: Based on the first position information, determine the projection reference point of the projected image on the target projection surface, generate corresponding interactive image data, and control the projector to project the interactive image data onto the target projection surface. S500: Activate the vision sensor, capture the action of the interactive object based on the region of interest determined by the first position information, update the interactive screen data according to the captured action, and control the interactive projection system to stop working when the interaction end condition is met, and control the robot to resume its mobility.

[0028] Step S100: Forced isolation between moving and non-working states. The main purpose is to eliminate the possibility of the interactive projection system interfering with the moving process during the entire process of the robot performing the moving task. More importantly, it is to prevent the projection image from being distorted due to the movement of the robot itself.

[0029] In practice, when the control unit (such as the main controller or navigation module) of the wheeled chassis robot receives a movement command (such as going to a specific exhibition area A) and starts driving the chassis motor to move, it immediately determines that it has entered the movement state.

[0030] The control unit synchronously sends a first control command to the interactive projection system integrated into the robot body. This command is not a simple request, but a mandatory status command, the core semantic of which is "moving".

[0031] Upon receiving the instruction, the control server of the interactive projection system will strictly execute the predetermined security policy to ensure that the entire system remains in a non-operating state. This non-operating state is not a simple software hibernation, but a hardware-level control. This invention eliminates the direct impact of vibration and swaying of the mobile platform on the projection light path, ensuring the potential quality of the subsequent projected image from the source, while also significantly reducing the system's energy consumption during non-essential operating phases.

[0032] Step S200: Point Readiness and State Transition Trigger. The main purpose is to accurately and reliably identify the transition point of the robot from a moving state to a stationary state that can be used for high-quality projection, and use this as the only legitimate trigger condition for activating interactive functions.

[0033] Once the robot determines that it has reached the vicinity of the target location using its own navigation system (such as SLAM), it will perform a stopping operation. However, simply stopping does not mean that it is ready for high-precision projection.

[0034] Therefore, a dedicated state determination step was implemented to confirm that the robot has entered a stable, stationary state. This determination effectively filters out momentary vibrations caused by uneven ground, braking inertia, etc., ensuring the robustness of the determination.

[0035] Once the determination is successful, the control unit immediately sends a second control command (such as "point-ready") to the interactive projection server. This command is mutually exclusive with the command from S100, unlocking all advanced functions of the projection system. This design establishes a clear and unambiguous state switching protocol, ensuring that the projection system will never be accidentally activated before the robot has reached a stable, stationary state.

[0036] Step S300: High-precision object localization in a static environment. The main purpose is to quickly and accurately obtain the precise spatial position of people (or other interactive objects) within the interactive area by using sensors specializing in distance measurement, based on the stable condition of the robot itself being stationary.

[0037] Once the control core of the interactive projection server responds to the positioning ready command from S200, it first activates the positioning sensor. Preferably, this positioning sensor is a LiDAR. The LiDAR is activated to full-power operating mode and performs high-speed rotational scanning of the preset interactive area around the robot to obtain two-dimensional or three-dimensional point cloud data of the environment.

[0038] Subsequently, the server executes the core data processing flow: clustering non-ground points, identifying laser points belonging to the same person as an independent point cloud cluster; for each identified point cloud cluster, calculating the two-dimensional coordinates of its bottom center point in the robot coordinate system, which is defined as the first position information.

[0039] For example, this coordinate can approximate the center point of a person's standing position. The acquired initial positional information is an extremely reliable and accurate spatial anchor point, providing a unique and authoritative geometric reference for all subsequent position-related operations (image projection, visual focusing), enabling precise interaction.

[0040] Step S400: Adaptive image generation and stable projection based on spatial anchor points. The main purpose is to accurately align and stably present virtual interactive content with real physical space, and to drive the entire projection process based on the first position information generated in the previous stage.

[0041] After obtaining the initial location information, the control server does not immediately project a fixed image. Instead, it first makes a decision: based on the preset scene configuration, it chooses whether to project the interactive image onto the ground or the wall directly in front of it.

[0042] After determining the reference point for ground / wall projection, the server generates corresponding interactive screen data (such as a circular menu centered on that point). Finally, the screen data stream is sent to the projector, which projects it onto the target surface. Since the robot chassis is in an absolutely stationary state as confirmed by S200, there is no relative movement between the projector's optical components and the target surface. Therefore, after the initial alignment, no complex real-time dynamic compensation is required, and the image maintains extremely high static stability throughout, with clarity indistinguishable from a fixed projection system.

[0043] Step S500: Focus on real-time interaction and system reset under perception. The main purpose is to achieve low-latency and high-accuracy human-computer interaction, and to safely and orderly return the system to the mobile standby state after the interaction is completed.

[0044] While the image is being projected stably, the control server activates the vision sensor (such as an RGB-D camera). To improve processing efficiency and reduce false recognition, the vision system does not process the entire image, but instead uses a focused perception strategy: the server maps the first position information obtained from the S300 onto the imaging plane of the vision sensor through coordinate transformation, and uses these coordinates as the center to define a preset area of ​​interest.

[0045] The visual algorithm performs operations such as skeletal keypoint detection and gesture recognition only within this area. Once a specific user action is captured, the control server generates corresponding visual feedback content in real time, integrates it into the existing interactive screen data, and drives the projector to refresh the projected image, thus creating instant interactive feedback.

[0046] By utilizing precise location prior information, the scope of visual processing is greatly narrowed, which not only significantly improves the real-time performance and computational efficiency of action recognition, but also effectively avoids misidentification caused by background interference, thus achieving accurate interaction.

[0047] When the preset interaction termination conditions are met, the control server will sequentially turn off the projector, switch the sensors back to low-power mode, and send a completion signal to the robot control unit. The control unit will then release the robot from its static lock state, restore its mobility, and the entire system will switch back to the mobile standby state before S100, thus forming a complete automated process that can be executed in a loop.

[0048] This invention not only systematically solves the stability problem of mobile projection, but also realizes high-quality, high-precision, and high-safety fixed-point interactive projection on mobile robot platforms, providing wheeled service robots with powerful environmental interaction capabilities.

[0049] As a preferred embodiment of the present invention, keeping the interactive projection system in a non-operating state specifically includes: Send an instruction containing a moving status identifier to the control server of the interactive projection system; In response to the instruction, the control server executes at least one of the following: Disconnect or disable power supply to the projector; switch the positioning sensor and / or the vision sensor to a low-power standby mode.

[0050] This implementation ensures the safety and energy efficiency of the wheeled chassis robot during movement, and eliminates the risk of a decrease in projection quality due to platform movement.

[0051] When the robot's control unit (such as the navigation master controller) determines that the robot has entered a moving state (for example, the chassis motors start to drive and the path planner is activated), its primary task is not just to control the movement, but to synchronously generate and send an instruction with a clear state identifier.

[0052] This instruction is typically structured as a data packet containing a key field, such as "System Status: Moving". This instruction is then directed to the control server of the interactive projection system integrated into the robot.

[0053] This establishes a clear, real-time, and unambiguous state communication protocol between the robot's different functional subsystems. It is a mandatory state declaration that ensures the control core of the projection system can maintain strict synchronization with the robot's global motion state, providing a basis for subsequent hardware-level control.

[0054] Upon receiving a "moving status" command from the robot control unit, the interactive projection system's control server immediately interrupts any ongoing or standby interactive tasks and triggers a pre-defined safety control process. This process is designed for the highest level of non-working, specifically by performing at least one of the following operations on critical hardware components: For power level control of the projector, the control server sends instructions to the projector's power management module to directly cut off the power supply to the projector's core light-emitting components such as the optical engine, laser, or bulb, or to prevent its high-voltage drive circuit from being powered on.

[0055] This is a complete hardware shutdown, equivalent to physically unplugging the projector. This strategy ensures that the projector's optical output is absolutely stopped, eliminating any possibility of accidental projection due to circuit afterglow, standby light leakage, or software malfunctions. It achieves zero projection output at the physical light source level, providing the highest level of safety.

[0056] To mitigate the power consumption and functional degradation of sensors, the control server synchronously manages both positioning and vision sensors. For positioning sensors (such as LiDAR), they are switched from high-frequency, high-precision scanning modes to low-power standby modes. In this mode, the LiDAR may stop rotating and emitting light, or perform single-point ranging at extremely low frequencies for minimal obstacle avoidance. Its data processing unit is suspended, significantly reducing power consumption and ensuring that it does not generate high-density point cloud data for interaction. For vision sensors, they are similarly switched to low-power standby modes, for example, by turning off the structured light emitter, reducing the image sensor frame rate to 1-5 frames per second, or running only simple motion detection algorithms for security monitoring, while disabling complex advanced functions such as skeletal recognition and gesture recognition.

[0057] This achieves functional isolation, meaning that although the sensor is not completely powered off, its core interactive sensing capabilities have been actively stripped away, retaining only the most basic sensing capabilities required to maintain mobile safety.

[0058] A deep isolation from logic to physical layer is established. Even if a logical error occurs in the upper-layer software, the hardware's power-off or degradation instructions can act as a last line of defense, ensuring that the interactive system cannot be erroneously activated during movement. This eliminates the safety hazards of visual misdirection, accidental operation, or even collisions that may be caused by projected image jitter or misalignment during movement. During this period, the high-power projector is turned off, and the high-computing-power-requirement sensors are operated at minimum power, which can significantly reduce the overall system energy consumption and extend the robot's battery life.

[0059] In a preferred embodiment of the present invention, determining whether the robot has entered a stationary state includes: Monitor the robot's wheel speed and attitude angle data; If the wheel speed remains zero and the change in the attitude angle is less than a first threshold within a preset time window, then the robot is determined to have entered the stationary state.

[0060] In this embodiment, determining whether the robot has entered a stationary state does not simply rely on a single event such as a stop command or a motor stop signal. Instead, a comprehensive and robust determination logic based on multi-sensor data fusion and duration verification is constructed to accurately and reliably capture the key state transition of the robot from dynamic motion to a static platform that can be used for stable projection.

[0061] The control unit receives and processes real-time data streams from two independent sensor subsystems in parallel: Wheel speed data monitoring continuously reads feedback signals from the wheel hub encoders or motor drivers to obtain the actual rotational speed of each drive wheel of the robot. This data directly reflects the translational motion state of the robot chassis. When the rotational speed of all drive wheels is zero, it indicates that, from the power output perspective, the robot has stopped its active translational movement.

[0062] Attitude angle data monitoring continuously reads attitude angle data provided by the inertial measurement unit (IMU), mainly including pitch and roll angles. The IMU uses gyroscope and accelerometer fusion calculations to sensitively perceive the robot's angular motion and minute vibrations in three-dimensional space. Attitude angle data reflects the robot's attitude stability caused by uneven ground, its own inertia, or external disturbances.

[0063] The control unit compares the aforementioned real-time monitoring data with a set of preset, stringent judgment criteria. These criteria together form a state filter to distinguish between two different states: having just stopped and being stably stationary. Condition 1 (Translational Stationary Condition): The wheel speed must remain zero. "Remaining" means that the wheel speed signal must remain at zero throughout the entire subsequent judgment time window. Any instantaneous non-zero pulse (such as that caused by ground slippage or control vibration) will cause the condition to be unsatisfactory.

[0064] Condition 2 (Attitude Stability Condition): The change in attitude angles is less than a first threshold. The control unit calculates the magnitude of change in pitch and roll angles (e.g., the difference between the maximum and minimum values, or the root mean square value) within the decision time window. This first threshold is set to a very small angle value (e.g., 0.5 degrees or 1 degree) to detect whether residual swaying of the robot body caused by suspension slack, center of gravity settling, or environmental vibrations has decayed to a negligible level.

[0065] The decision is not based on instantaneous sampling, but rather on continuous logical verification of the two conditions within a preset time window (e.g., 1.5 to 3 seconds). Only when both conditions one and two remain true throughout the entire time window does the control unit ultimately decide to enter the static state. Once the decision is made, the control unit immediately generates and sends a point-ready command to the interactive projection server.

[0066] By simultaneously verifying translational stillness and attitude stability, this method ensures that the robot is not only stationary in position but also sufficiently stable in attitude within three-dimensional space. Even minor wobbling of the projector's mounting base (i.e., the robot itself) can be amplified into noticeable jitter in the projected image through leverage. This method, by setting a strict attitude change threshold, ensures that when projection begins, the robot's wobbling has attenuated to a level sufficient to support the demanding application of static projection, thus providing a physical guarantee for a clear and stable projected image.

[0067] Furthermore, the transition of a robot from motion to complete stillness is a dynamic process, accompanied by transient responses such as braking inertia and suspension oscillations. A simple method of triggering the projection at zero wheel speed is highly susceptible to prematurely activating it before the robot has truly stabilized, resulting in an initial, shaky image. The introduction of a preset time window forces the system to observe a complete decay cycle, effectively filtering out these brief transitional states and potential sensor noise interference. This significantly improves the accuracy and reliability of state determination, preventing premature or unstable activation of the projection system due to misjudgment.

[0068] Specifically, after determining that the robot has entered the stationary state, the process further includes: Trigger the robot's braking device or telescopic support legs to enhance static stability.

[0069] Once the control unit confirms the stationary state, it will immediately generate and send control commands to trigger the robot's braking device or telescopic support legs, thereby further enhancing the robot's static stability at the target point from a mechanical perspective.

[0070] The control unit sends a full brake lock command to the electronic braking module of the chassis drive system. This command not only requires the motor to stop (already achieved in the stationary determination), but also requires the application of an active holding torque to the motor shaft or the triggering of an independent disc brake / holding brake mechanism.

[0071] The braking module responds to commands by applying electrical or mechanical braking to the drive wheels. For electrical braking, a specific current may be injected into the motor windings via the driver to generate holding torque, locking the axle. For mechanical braking, electromagnetic brake pads are activated. The system may optionally monitor braking pressure or current feedback to confirm that braking has been reliably applied. This strategy aims to eliminate any slight rolling or wobbling tendencies that may arise in the wheel train due to gear backlash, bearing clearance, or external forces, transforming the mobile chassis into a platform locked in translational freedom.

[0072] The control unit sends extension commands to the support leg drive mechanism (such as a motor or cylinder). Typically, the robot's chassis has three or four retractable support legs (e.g., driven by screws, links, or pneumatic actuators).

[0073] The support legs extend from the bottom of the robot's housing until their feet establish stable physical contact with the ground. In a preferred embodiment, the support legs may be designed to have a certain force sensing or stroke sensing capability, and stop extending when they detect that they have made contact with the ground and are subjected to a preset small pressure (indicating that force has been applied but the robot body has not been excessively lifted).

[0074] Once multiple support legs are in contact with the ground and working together to bear weight, they, together with the robot chassis, form a multi-point rigid support structure. This structure greatly improves the overall structural rigidity of the robot, effectively suppressing all minute movements in the six degrees of freedom (three translations and three rotations) caused by the chassis suspension system, tire deformation, or localized soft ground, thus securing the robot with rigid support legs.

[0075] By actively intervening mechanically, any residual micro-vibrations, minor subsidence or deformation caused by the ground material (such as soft carpet), and potential minor external disturbances (such as airflow caused by people walking or vibration transmitted through the ground) can be offset or eliminated, thereby creating an extremely stable and near-ideal rigid reference plane for the projection light path and visual perception.

[0076] This ensures that the projected image is free from blurring or jitter caused by platform movement, improving static clarity and achieving a high-quality, highly immersive interactive experience.

[0077] As a preferred embodiment of the present invention, determining the first position information of at least one interactive object in the robot coordinate system based on scan data includes: The positioning sensor is a lidar; The point cloud data acquired by the lidar is subjected to ground filtering and clustering processing to identify point cloud clusters corresponding to the interactive objects; Calculate the two-dimensional coordinates of the bottom center point of the point cloud cluster in the robot coordinate system, and use it as the first position information.

[0078] In a preferred embodiment of the present invention, the positioning sensor is preferably a lidar. Lidar emits a laser beam by rotating at high speed and receives reflected signals, enabling it to acquire high-density, high-precision distance information about the robot's surrounding environment, forming three-dimensional point cloud data. The purpose is to transform complex physical environment data into accurate spatial geometric information that the system can understand and use.

[0079] Once the robot reaches a stable stationary state and receives the positioning ready command, the control core of the interactive projection server immediately activates the LiDAR. The LiDAR performs a 360-degree or sector scan within a pre-defined interactive area (e.g., a 120-degree sector in front of the robot with a radius of 5 meters), acquiring tens of thousands of point cloud data points in a very short time. Each data point contains its distance, horizontal, and vertical angle information in the LiDAR's own coordinate system. The server first converts this raw data to the robot's coordinate system to facilitate subsequent processing and alignment with the coordinates of other robot modules (such as the projector).

[0080] Point cloud data contains information about all objects, including the ground, walls, furniture, and people. To efficiently identify people, the first step is to separate the ground point cloud. The server uses a plane fitting algorithm (such as RANSAC) or a method based on a pre-defined ground model to identify and filter out point clouds that conform to the features of the ground plane.

[0081] Specifically, the algorithm searches for an optimal planar model in the robot coordinate system (usually close to the z=0 plane, allowing for a certain tilt tolerance), and classifies points falling within a certain threshold range near this plane as ground points and removes them. This step greatly reduces the amount of data processed subsequently and eliminates the biggest source of interference—the ground—making non-ground objects (mainly standing people) stand out in the point cloud.

[0082] After filtering out the ground, the remaining point cloud represents all obstacles and interactive objects in the environment. Next, clustering algorithms (such as the DBSCAN algorithm based on Euclidean distance or methods based on connected components) are used to group these discrete points. The principle of the algorithm is that points belonging to the same physical entity (such as a person) are close in three-dimensional space, while points belonging to different entities have significant distance intervals. By setting an appropriate neighborhood search radius and a minimum point count threshold, the algorithm can automatically aggregate point clouds representing the same person into independent point cloud clusters. Each identified point cloud cluster is then preliminarily determined to be a potential interactive object.

[0083] For each identified point cloud cluster of interactive objects, a stable and geometrically meaningful point needs to be calculated to represent its standing position. This implementation uses the bottom center point of the point cloud cluster as the first position information. Specifically: Once the bottom is determined, in the robot coordinate system, find the lowest height value (i.e., the minimum Z coordinate value) of all points in the cloud cluster, or take all points with height values ​​within a certain low range. These points usually correspond to the contact area between the human's feet or lower legs and the ground.

[0084] Calculate the center: For these bottom points, calculate their geometric center in the horizontal plane (XY plane). This can be obtained by calculating the average of the X and Y coordinates of these bottom points.

[0085] The output coordinates, and the resulting two-dimensional coordinates (X_center, Y_center), are defined as the first position information of the interactive object in the robot coordinate system. These coordinates approximately represent the center of the person's standing position and are a stable and reliable physical reference point.

[0086] The distance information provided by lidar has absolute accuracy at the centimeter or even millimeter level, and is unaffected by changes in ambient lighting (such as strong light, weak light, or backlight), and is completely independent of the content of the projected image. This makes the calculated initial position information an accurate and reliable spatial anchor point based on physical measurements.

[0087] Through two key processing steps—ground filtering and clustering—this method can reliably separate and identify standing people from the background (ground, walls, static furniture) in complex indoor environments. This method is insensitive to a person's posture or clothing; as long as their body outline can be scanned by LiDAR, they can be effectively identified, demonstrating strong robustness.

[0088] Choosing the bottom center point as location information has clear physical significance and practicality. It directly corresponds to the center of contact between a person and the ground, serving as the base point for the natural placement of the interactive interface. Compared to using the highest point, center of gravity, or outer contour center of the point cloud, the bottom center point is least affected by upper body movements such as waving or turning, making it a very stable reference point whether the person is standing still or moving within a small range.

[0089] In a preferred embodiment of the present invention, controlling the projector to project the interactive image data onto the target projection surface includes: According to the preset scene configuration, the target projection surface is selected as the ground or the wall. If the ground is selected, the ground point corresponding to the first location information is used as the projection reference point; If a wall is selected, the vertical projection point of the first position information on the wall is calculated based on the robot's current position and orientation, and used as the projection reference point. Based on the pre-calibrated projector installation parameters and coordinate system transformation relationship, the interactive screen data is geometrically corrected to compensate for the screen distortion caused by the tilted installation of the projector.

[0090] The goal of this implementation is to project dynamically generated interactive images onto a selected physical surface with high quality, according to the specific needs of the scenario.

[0091] The control server does not project the image onto a single surface in a fixed manner, but rather dynamically selects the projection surface based on preset scene configurations (e.g., instruction files, interactive content types, or environmental parameters). The ground projection option is suitable for scenarios such as user foot interaction, ground games, and navigation instructions. For example, when the interactive content is a selection menu surrounding the user or a soccer game, choosing the ground as the projection surface is the most natural.

[0092] Wall projection is suitable for information displays, large-screen interactions (such as virtual touchscreens), or scenarios where the floor is unsuitable for projection (such as reflective materials or areas with frequent foot traffic). For example, projecting a product introduction screen or a virtual keyboard for user operation.

[0093] This decision-making process gives the system a high degree of scene adaptability, allowing the same robot to flexibly adapt the interactive interface to the most suitable physical medium according to different tasks.

[0094] After determining the projection surface, the core task is to determine the precise anchor point of the interactive image on the physical surface, i.e., the projection reference point. This calculation uses the initial position information obtained from the lidar (i.e., the center coordinates of the interactive object's standing point) as the core input: Ground reference point calculation (direct mapping): When a ground location is selected, the control server directly uses the initial location information as the reference point. This is equivalent to defining a reference origin on the ground near the user's standing point, and subsequent screen content will be arranged with this origin as the center or reference.

[0095] Wall reference point calculation (projective geometric transformation): When a wall is selected, the control server needs to perform geometric calculations by combining the robot's current position and orientation in the global coordinate system (or in the room map) with the pre-stored or real-time detected wall space plane equations.

[0096] The specific method is as follows: From the first position information point in the robot's coordinate system, draw a straight line perpendicular to the ground (representing the standing vertical line of the human body), and then calculate the intersection point of this line with the spatial plane where the wall is located. This intersection point is the vertical projection point of the person's standing position on the wall. This point will be used as the reference point for the wall projection. This ensures that the interactive screen (such as a virtual screen) is facing the user and has a clear directional relationship with the user.

[0097] In practical systems, for aesthetic reasons, compact design, or to avoid obstruction, projectors are typically not installed directly facing the target surface (floor or wall), but rather at a certain angle. Without proper handling, this can result in severe trapezoidal distortion of the projected image; for example, a square image will become a trapezoid. Therefore, this step is crucial for ensuring the correct geometry of the image.

[0098] Pre-calibration: A one-time projector calibration is performed before system deployment or at the factory. By acquiring images of a specific pattern (such as a grid) projected onto a target plane by the projector, and combining this with analysis of pattern deformation using the robot's sensors (such as vision sensors or additional calibration cameras), the installation pose of the projector in the robot coordinate system (i.e., position and three rotation angles) is accurately calculated, and a projection transformation model of the pixel coordinates of the projector image plane and the three-dimensional coordinates of the robot space is established.

[0099] Real-time correction: During each projection, the control server uses the calibrated transformation model and the projection reference point determined for this time to perform a reverse geometric transformation on the original interactive screen data (designed on an ideal facing plane).

[0100] Specifically, it calculates the new coordinates in the projector's image frame buffer for each pixel in the original image, taking into account the projector's tilt, the reference point's position, and the target surface's plane equation. Then, it redraws the image content onto the projector's input image according to this mapping. This process is commonly called keystone correction or projection transformation. The result is that even though the projector is installed at an angle, the shape of the image projected onto the floor or wall is completely consistent with the original design, achieving a distortion-free and accurate presentation.

[0101] By using primary location information as the basis for calculations, the interactive screen is ensured to always focus on the user and serve as the frame of reference, creating a natural interactive experience where the interface proactively adapts to the user. Whether it's a menu on the floor or a screen on the wall, users can intuitively perceive its spatial relationship with themselves, making the interaction intent clear without requiring users to move to adapt to a fixed interface position.

[0102] Through real-time, precision-calibrated geometric correction, image distortion caused by equipment installation limitations is completely eliminated. Projected circles are perfect circles, squares are perfectly straight squares, and text is clear and unstretched. This ensures accurate transmission of visual information and an aesthetically pleasing experience, improving the projection quality of mobile robots.

[0103] In a preferred embodiment of the present invention, a visual sensor is activated to capture the action of the interactive object based on the region of interest determined by the first location information, and the interactive screen data is updated according to the captured action.

[0104] The region of interest determined based on the first location information refers to a preset range centered on the projection coordinates of the first location information on the imaging plane of the visual sensor.

[0105] The core objective of this step is to utilize the high-precision initial position information obtained from positioning sensors such as LiDAR to define a highly relevant and spatially defined processing area for motion capture by visual sensors (such as RGB-D cameras). This aims to address the problems faced by general-purpose visual algorithms when searching for human bodies or actions across the entire image range, including high computational overhead, significant background interference, and limited real-time performance.

[0106] The interactive projection server first needs to accurately map the initial position information (X_center, Y_center) from the LiDAR coordinate system (or a unified robot coordinate system) to the image coordinate system of the vision sensor. This process relies on pre-completed joint calibration between the two sensors. Using the rotation matrix and translation vector (extrinsic parameter matrix) obtained through calibration, the server can project the initial position information point in three-dimensional space onto the two-dimensional imaging plane of the vision sensor, obtaining a pixel coordinate (u_center, v_center). This coordinate point represents the theoretical projection position of the center of the person's standing point in the visual image.

[0107] After obtaining the center pixel coordinates (u_center, v_center), the server does not process this single point in isolation. Instead, it dynamically defines a preset area, such as a rectangle, circle, or other shape, centered on this point as the region of interest for visual processing. The size of this preset area is an adjustable parameter of the algorithm, and its design strategy is as follows: The area should be large enough to accommodate a person’s typical activity space (e.g., from the feet to the head, and the range of arm extension), ensuring that most of the user’s expected movements (such as waving, kicking, bending over) fall within the area.

[0108] The area should not be too large to avoid including too much irrelevant background and reducing processing efficiency; nor should it be too small to prevent minor user movements or small errors in projection coordinate estimation from causing the body parts to move out of the ROI.

[0109] Typically, this range might be a rectangle with sides of 150-300 pixels (depending on camera resolution and distance).

[0110] Taking a rectangular region of interest (ROI) as an example, the coordinates of the top-left corner (u_left, v_top) and the bottom-right corner (u_right, v_bottom) of the ROI can be determined in the following way: u_left = u_center - W / 2, v_top = v_center - H / 2, u_right = u_center + W / 2, v_bottom = v_center + H / 2, Where W and H are the preset ROI width and height, respectively.

[0111] Furthermore, the ROI can be dynamically scaled based on the estimated distance to people (obtainable from the Z-coordinate of the first location information or the visual depth map). The closer people are, the larger the ROI; the farther people are, the smaller the ROI, to always maintain a reasonable proportion of human body coverage.

[0112] After defining the Region of Interest (ROI), the server sends this region information (i.e., pixel coordinate boundaries) to the vision processing module. Upon receiving each frame, the vision module performs advanced processing only on the pixels within the ROI. For the image portion outside the ROI, the vision module does not need to perform the aforementioned complex calculations and may only perform low-complexity background subtraction or simply ignore it.

[0113] By limiting computationally intensive vision algorithms to only a small percentage of the total... Figure 1 Within a small portion (typically <20%) of the ROI, the computational load, memory access requirements, and time overhead for processing each frame are significantly reduced. This allows the system to achieve higher frame rates for motion capture and lower interaction latency with limited hardware computing power, ensuring a smooth user experience.

[0114] Furthermore, by focusing the processing area around the user, most of the interference from irrelevant background information is physically eliminated, providing a clean input environment for the visual algorithm, thereby significantly improving the accuracy and reliability of motion capture.

[0115] The updating of the interactive screen data refers to the control server generating corresponding visual feedback content in real time based on the action type identified by the visual sensor, and integrating it into the interactive screen data to control the projector to update the projected screen.

[0116] The core objective of this step is to establish a low-latency, highly coupled mapping system. This system enables the rapid recognition of every user interaction (such as clicking, waving, or jumping) and instantly drives corresponding, visually appealing dynamic changes in the projected image. This real-time feedback mechanism not only confirms to the user that their action has been perceived by the system (providing a sense of confirmation), but more importantly, it guides the user to the next interaction through the dynamic evolution of visual content, thus forming a smooth and coherent interactive narrative.

[0117] Within the defined Region of Interest (ROI), the vision processing module continuously runs the action recognition algorithm. When a pre-defined, valid interactive action is detected (e.g., "extending an arm forward" is defined as a "select / click" action, and "waving to the left" is defined as a "page turning" action), the module does not simply output a category label, but generates a structured interaction event. This event data packet contains at least the following information: Action type identifier, a code that uniquely identifies this interactive action; Trigger timestamps to precisely record the moment an action occurs; Action space information, optionally, may include the approximate location where the action occurs (e.g., relative coordinates within the ROI) for interactions that require spatial alignment (such as clicking a virtual button).

[0118] The event data packet is sent to the control server of the interactive projection system in real time.

[0119] The control server internally maintains the state machine and interaction logic model of the current interactive scene. When it receives an interaction event from the vision module, it calls or generates corresponding visual feedback content in real time based on the action type in the event and the current system state. This content is not a pre-rendered complete video, but often includes: Dynamic graphic elements, such as a glowing button animation, a ripple effect, or a trajectory line that follows a gesture.

[0120] Status switching elements, for example, the highlighted items in a menu interface switch from one to another, or the 3D model of a virtual item changes from gray to color.

[0121] The server can also send commands to the robot's audio module to trigger corresponding sound effects, thus achieving synesthetic feedback.

[0122] The design of feedback content is deeply integrated with the interaction logic, emphasizing immediacy, expressiveness, and logical clarity.

[0123] After generating the feedback content, the control server needs to seamlessly integrate it into the currently projected interactive screen data. This interactive screen data forms the base layer of the entire interactive interface (such as fixed menu backgrounds or game scenes). Integration is not a simple replacement, but a real-time compositing process based on graphic overlay or pixel blending. The server first renders the base screen in a graphics rendering buffer, and then overlays dynamic effects on top of it according to the requirements of the feedback content. This requires the server to have a certain level of real-time graphics rendering capabilities.

[0124] After the new frame of image data is synthesized, the control server sends the updated entire image data (or only the changed areas) to the projector. The projector then immediately refreshes its output, projecting the image, which incorporates the latest visual feedback, onto the target surface. Because the entire processing chain from motion recognition to image update is highly optimized, and the computation is limited to the ROI, the feedback latency can be controlled to an extremely low level. Users can hardly perceive the lag between operation and image changes, thus obtaining a smooth interactive experience.

[0125] By translating users' physical actions into visual changes in their surroundings in real time and intuitively, environmental responsiveness is achieved, creating an immersive experience. Real-time visual feedback provides users with immediate operational feedback. When a user performs an action, they can immediately see the corresponding change on the projected screen, enhancing their sense of control and confidence. Simultaneously, the feedback content naturally guides the user to the next step, making the interaction flow smooth and natural.

[0126] By employing a strategy of seamless updates rather than full-screen transitions, the visual continuity of the interactive interface is ensured. Users remain in a stable visual environment, with only the parts directly related to the interaction undergoing graceful evolution. This avoids abrupt screen jumps, reduces the user's cognitive load, and enhances the comfort and professionalism of the interaction.

[0127] In a preferred embodiment of the present invention, the interaction termination condition includes any one of the following: The preset time has elapsed since the interactive projection was initiated; The visual sensor recognizes the preset end gesture or command; The robot receives a new movement task instruction sent from the outside.

[0128] The core objective of this step is to establish a diversified, robust, and intuitive exit mechanism for the resource-intensive task of point-to-point interactive projection. This mechanism needs to balance multiple dimensions of requirements: ensuring the system is not overwhelmed by a single task (resource management), respecting the user's interaction intentions (natural interaction), and responding to higher-priority task scheduling (system collaboration). By pre-setting multiple termination conditions, the system can intelligently determine when to terminate the current interaction and execute a standardized shutdown procedure to ensure the projection system is safely shut down, the robot regains its mobility, and prepares for the next task cycle.

[0129] Automatic termination based on preset duration (Condition 1, timeout protection): When the interactive projection system receives the point-ready command and officially starts all sensors and projectors, a countdown timer is started internally by the control server (or robot main control unit). The duration of this timer is set to a preset value, such as 3 minutes, 5 minutes, or a specific time configured according to the interactive content. This duration is usually set based on the longest time required for a typical single interaction, leaving a reasonable margin.

[0130] When the countdown timer reaches zero and other termination conditions have not yet been triggered, the system determines that the preset duration termination condition has been met. At this point, the control server will generate an interaction termination event and initiate a standardized shutdown sequence, gradually shutting down the projector, switching the sensors to standby mode, and notifying the chassis control system to release the stationary lock.

[0131] This strategy serves as a safety net and resource recycling mechanism, effectively preventing robots from remaining in one place for extended periods, continuously consuming energy, and extending equipment lifespan due to users leaving without giving a clear end signal or system software malfunctions (such as failing to recognize the end gesture).

[0132] User-initiated termination based on visual recognition (Condition 2, natural interaction termination): During the interaction, the visual sensor continuously runs a set of gesture or posture recognition algorithms specifically designed for ending the interaction within a defined area of ​​interest. The system pre-sets one or more clear, easy-to-execute termination gestures or commands that are not easily confused with regular interaction actions. For example, the user can cross their arms in front of their chest to form an "X" shape, wave twice consecutively, make a clear "goodbye" gesture, or select a virtual "exit" button on the interactive interface (by clicking through motion capture).

[0133] When the visual processing module recognizes a preset end gesture or command, it immediately sends a specific "user end" event to the control server. The control server prioritizes the user's active intention, immediately terminating the current interaction process and initiating a standardized shutdown sequence.

[0134] This strategy empowers users with the most direct and natural control over the interaction, aligning with a human-centered interaction logic where the interaction begins with the robot's arrival and ends with the user's gesture. It respects the user's desire to end the interaction, providing a smooth, contactless exit experience and enhancing user-friendliness.

[0135] Forced termination based on external task scheduling (condition 3, task priority response): The robot's control unit (master controller) continuously listens for external command channels, such as wireless commands from the central scheduling system, collaboration requests from other robots, or remote control commands from the operator. These commands may include new movement task commands with higher or equal priority, such as "immediately proceed to area B to perform an emergency task", "return to the charging station", or "provide service to another user".

[0136] Once the control unit receives such a new movement task instruction, regardless of whether the current interaction is in progress or has timed out, it immediately sends a high-priority "interrupt" instruction to the interactive projection server, or directly treats it as having met the interaction termination conditions. The control server must immediately interrupt the current rendering and interactive logic processing, and execute the shutdown sequence as quickly and safely as possible to release system resources and prepare for movement.

[0137] This strategy ensures the flexibility and schedulability of the robot as a holistic system asset. It prevents the robot from becoming bogged down in a single, lengthy interactive task and thus unable to respond to more urgent or global task demands, guaranteeing the robot's efficiency and collaborative capabilities in dynamic and ever-changing multitasking environments.

[0138] The system provides users with multiple exit paths, from natural gestures to automatic system timeout, covering various real-world scenarios such as user-initiated termination, involuntary forgetting, and unexpected interruptions. This design makes the system's behavior predictable and logical, reducing user confusion and learning costs.

[0139] The triple termination condition constitutes a redundant and complementary state exit guarantee system. The failure of a single condition (such as a temporary malfunction of the gesture recognition algorithm) will not cause the system to freeze, and other conditions (timeout or external command) can still ensure that the system eventually recovers to normal. This significantly improves the long-term operational reliability of the entire system in unpredictable real-world environments.

[0140] This invention also provides a wheeled chassis robot, such as Figure 2 As shown, it includes: Mobile chassis 1, used to provide the robot with mobility; Control unit 3 is used to control the movement of the mobile chassis 1; Interactive projection system 2, integrated into the robot body, includes positioning sensor 21, vision sensor 22, projector 23 and control server 24; The control unit 3 is configured to perform the method described.

[0141] Therefore, any effect that can be achieved in the fixed-point interactive projection control method for wheeled chassis robots will not be elaborated here.

[0142] For any parts not mentioned in this invention, existing technologies can be used or referenced.

[0143] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

[0144] The above description is merely an embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principle of the present invention should be included within the scope of the claims of the present invention.

Claims

1. A fixed-point interactive projection control method for a wheeled chassis robot, characterized in that, include: Control the wheeled chassis robot to move to the target position. When the robot is in the moving state, send a first control command to the interactive projection system integrated into the robot body so that the interactive projection system remains in a non-working state. Once the robot reaches the target location, it is determined whether the robot has entered a stationary state. If it is determined that the robot has entered the stationary state, a second control command is sent to the interactive projection system. In response to the second control command, the interactive projection system activates the positioning sensor to scan the interactive area around the robot, and determines the first position information of at least one interactive object in the robot coordinate system based on the scan data. Based on the first position information, the projection reference point of the projected image on the target projection surface is determined, and the corresponding interactive image data is generated. The projector is then controlled to project the interactive image data onto the target projection surface. The visual sensor is activated to capture the movements of the interactive object based on the region of interest determined by the first location information. The interactive screen data is updated according to the captured movements. When the interaction ends, the interactive projection system is controlled to stop working, and the robot is controlled to resume its mobility.

2. The method according to claim 1, characterized in that, To keep the interactive projection system in a non-operating state, specifically including: Send an instruction containing a moving status identifier to the control server of the interactive projection system; In response to the instruction, the control server executes at least one of the following: Disconnect or disable power supply to the projector; switch the positioning sensor and / or the vision sensor to a low-power standby mode.

3. The method according to claim 1, characterized in that, Determining whether the robot has entered a stationary state includes: Monitor the robot's wheel speed and attitude angle data; If the wheel speed remains zero and the change in the attitude angle is less than a first threshold within a preset time window, then the robot is determined to have entered the stationary state.

4. The method according to claim 3, characterized in that, After determining that the robot has entered the stationary state, the process further includes: Trigger the robot's braking device or telescopic support legs to enhance static stability.

5. The method according to claim 1, characterized in that, Determining the first position information of at least one interactive object in the robot coordinate system based on scan data includes: The positioning sensor is a lidar; The point cloud data acquired by the lidar is subjected to ground filtering and clustering processing to identify point cloud clusters corresponding to the interactive objects; Calculate the two-dimensional coordinates of the bottom center point of the point cloud cluster in the robot coordinate system, and use it as the first position information.

6. The method according to claim 1, characterized in that, Controlling the projector to project the interactive image data onto the target projection surface includes: According to the preset scene configuration, the target projection surface is selected as the ground or the wall. If the ground is selected, the ground point corresponding to the first location information is used as the projection reference point; If a wall is selected, the vertical projection point of the first position information on the wall is calculated based on the robot's current position and orientation, and used as the projection reference point. Based on the pre-calibrated projector installation parameters and coordinate system transformation relationship, the interactive screen data is geometrically corrected to compensate for the screen distortion caused by the tilted installation of the projector.

7. The method according to claim 1 or 5, characterized in that, The region of interest determined based on the first location information refers to a preset range centered on the projection coordinates of the first location information on the imaging plane of the visual sensor.

8. The method according to claim 1 or 7, characterized in that, The updating of the interactive screen data refers to the control server generating corresponding visual feedback content in real time based on the action type identified by the visual sensor, and integrating it into the interactive screen data to control the projector to update the projected screen.

9. The method according to claim 1, characterized in that, The interaction termination condition includes any of the following: The preset time has elapsed since the interactive projection was initiated; The visual sensor recognizes the preset end gesture or command; The robot receives a new movement task instruction sent from the outside.

10. A wheeled chassis robot, characterized in that, include: A mobile chassis is used to provide the robot with mobility. The control unit is used to control the movement of the mobile chassis; An interactive projection system is integrated into the robot body, including positioning sensors, vision sensors, a projector, and a control server; The control unit is configured to perform the method of any one of claims 1 to 9.

Images