Autonomous navigation control system of cabin cleaning machine in limited space

Abstract

The invention relates to an industrial control system, and discloses an autonomous navigation control system for a cabin cleaning machine in a confined space, and the system comprises a spatial frequency analysis module which is used for extracting fundamental wave features and phase offset of a periodic structural member in the confined space; the deviation compensation module is used for identifying the boundary of the structural member and generating a motion compensation instruction; the phase inertia deduction module is used for executing phase integral accumulation by utilizing the updated phase angle speed parameter and the running speed to maintain the continuity of a control clock when the signal-to-noise ratio is lower than a threshold value; and the slip suppression module is used for calculating a slip speed observation value and generating a negative feedback torque instruction when the slip speed observation value exceeds the limit so as to suppress slip. The problem of navigation instability of the cabin cleaning machine under the surface with strong interference, discontinuous structure and low adhesion coefficient is solved.

Description

Technical Field

[0001] This invention relates to an autonomous navigation and control system for a cabin cleaning machine in a confined space, belonging to the field of industrial control system technology. Background Technology

[0002] A cargo hold cleaning robot is an industrial mobile robot specifically designed for cleaning the interiors of large, enclosed or semi-enclosed compartments, commonly used for cleaning cargo holds and oil tanks on large cargo ships. Currently, in confined space cleaning operations on large cargo holds and oil tanks, mobile robots need to perform long-distance wall-hugging operations in narrow, semi-enclosed environments with highly repetitive structural features. In this scenario, the robot control system must maintain a constant distance from the cargo wall and cope with multiple interferences from high-concentration water mist, residual mud, and irregular obstacles at the bottom of the hold. Currently, most mainstream navigation solutions utilize real-time localization and mapping (RTL) technology for open outdoor environments, relying on LiDAR or visual sensors to collect environmental feature points and using probabilistic matching algorithms to correct the robot's pose in real time on a global map. This approach has proven mature in feature-rich and highly differentiated unconfined environments and has become the standard method for automated operations. However, applying this general navigation paradigm to the bottom of cargo holds... When dredging specific industrial scenarios, severe adaptability challenges are faced. The environment at the bottom of the hull typically exhibits a corridor effect, and the internal reinforcing ribs and rib structures have high geometric repeatability, resulting in a lack of uniqueness in the spatial distribution of feature points. When general real-time localization and mapping algorithms process such data, they are prone to matching confusion due to the high similarity of local features, causing jumps and accumulated errors in the robot's longitudinal position estimation. In addition, the working surface at the bottom of the hull is covered with oil, mud, and cleaning waste liquid, forming a slippery surface with an extremely low adhesion coefficient. This causes a serious deviation between the data calculated based on the wheeled odometer and the actual displacement. The dual interference of feature confusion at the perception layer and slippage mismatch at the execution layer makes it difficult for traditional control systems to maintain the stability and trajectory accuracy of long-distance operation.

[0003] Besides hardware adaptability limitations, existing conventional control software architectures also suffer from logical defects in closed environments. The lack of an external absolute reference point makes traditional navigation strategies inconsistent. For example, Chinese invention patent CN116080837A discloses an autonomous navigation and monitoring unmanned vessel for waterways, constructing a system that relies on GNSS positioning components to obtain absolute position information for autonomous navigation and obstacle avoidance. While this technology performs well in open water, if transplanted to a fully enclosed metal cargo hold or oil tank environment, the electrostatic shielding effect of the hull steel plates blocks satellite positioning signals, causing the core navigation logic to fail. Furthermore, based on the design of a stable water surface, it lacks a nonlinear sliding dynamic compensation mechanism for high-viscosity sludge surfaces, particularly at the characteristic height of the hull bottom. In repetitive operating conditions where external absolute coordinates cannot be obtained, simple obstacle avoidance logic alone cannot maintain the continuity of long-distance wall-hugging operation trajectory and control stiffness. To address these issues, the industry has attempted to introduce high-precision inertial measurement units or Doppler velocimetry radar to assist positioning, trying to improve system robustness by increasing sensor redundancy. While this has alleviated positioning drift to some extent, it has increased system hardware costs and computational load, and has not fundamentally solved the problem of control continuity when environmental features are missing or signals are interrupted. In particular, under extreme conditions such as temporary loss of external observation signals due to bulkhead damage, structural obstruction, or strong water mist interference, existing systems are forced to interrupt operations or execute emergency stop logic due to the loss of reference points, which cannot meet the stringent requirements of continuous, efficient, and automated operations in industrial settings.

[0004] Therefore, the technical problem to be solved by this invention is how to construct a deterministic control mechanism to cope with complex working conditions such as feature repetition, surface slipperiness and signal interruption without relying on expensive external auxiliary equipment, by utilizing existing environmental structural constraints and body feedback information. Summary of the Invention

[0005] To address the problems mentioned in the background art, the technical solution of this invention is as follows: An autonomous navigation and control system for a confined space cleaning machine, capable of handling cleaning operations in confined spaces under complex conditions such as repetitive internal structural features and slippery surfaces; comprising: a spatial frequency analysis module, used to acquire horizontal ranging data sequences from the working side of the cleaning machine, perform spatial frequency analysis on the horizontal ranging data sequences to extract the fundamental wave characteristics and current phase offset of periodic structural components within the confined space; a deviation compensation module, used to identify the boundaries of periodic structural components based on the fundamental wave characteristics, calculate the normalized progress value of the cleaning machine within the current structural cycle in combination with the phase offset, and generate motion compensation commands for the cleaning machine chassis based on the horizontal ranging data sequences; and a phase inertial deduction module, used to calculate the signal-to-noise ratio of the fundamental wave characteristics in real time. The system employs several mechanisms: When the signal-to-noise ratio (SNR) exceeds a preset threshold, the phase angular velocity parameter is updated based on continuously acquired phase offsets. When the SNR falls below a confidence threshold, the output of the spatial frequency analysis module is masked, and the phase offset is integrated and accumulated using the phase angular velocity parameter and the real-time travel speed of the cleaning machine to maintain control clock continuity. A slip suppression module performs a time-dependent differential operation on the phase offset and generates a structural feed speed based on preset spacing parameters of periodic structural components. The structural feed speed is then differentially calculated with respect to the theoretical linear velocity of the cleaning machine's traveling mechanism to obtain the observed slip speed. When the observed slip speed exceeds a preset slip threshold, a negative feedback torque command is generated and superimposed on the drive control loop of the cleaning machine's traveling mechanism to reduce the deviation between the theoretical linear velocity and the structural feed speed.

[0006] Preferably, the spatial frequency analysis module is specifically used to convert the time-domain horizontal ranging data sequence into a frequency-domain signal, identify the specific frequency band with the highest amplitude as the fundamental frequency, confirm the data point corresponding to the fundamental frequency as the position of the periodic structural component, and calculate the phase difference between the wave peak signal in the horizontal ranging data sequence and the current position of the cleaning machine as the phase offset.

[0007] Preferably, the phase inertial deduction module is also used to execute smooth transition logic. When the signal-to-noise ratio index recovers from below the confidence threshold to above the confidence threshold, the phase deviation value between the deduced phase obtained by integral accumulation and the re-extracted measured phase is calculated. The phase deviation value is gradually reduced within a preset transition period by a weighted average algorithm until the system state is completely switched to the locking mode based on measured features.

[0008] Preferably, the deviation compensation module further includes a gain scheduling unit, which is used to calculate the normalized progress value of the cleaning machine in the current structural cycle based on the phase offset, and increase the weight of angle deviation control when the normalized progress value indicates that the cleaning machine is approaching the boundary of the periodic structural component, and increase the weight of lateral distance deviation control when the normalized progress value indicates that the cleaning machine is located in the middle area of ​​the periodic structural component.

[0009] Preferably, the slip suppression module follows the following calculation rules when generating the structural feed rate: ,in, Defined as the structural feed rate. Defined as the rate of change of the differential of the phase offset with respect to time. Defined as the preset spacing parameter for periodic structural components. Pi is a constant; the slip suppression module uses this calculation rule to map the rate of change of the phase offset to the absolute linear velocity of the cleaning machine relative to the physical boundary of the confined space.

[0010] Preferably, the slip suppression module is also used to perform torque redistribution operation. When the slip speed observation value exceeds the slip threshold, the conventional speed control command is bypassed, and the torque output of the slipping drive wheel is directly reduced or the torque difference between the left and right drive wheels is adjusted until the slip speed observation value converges to within the slip threshold, thereby realizing closed-loop control of heterogeneous speed field without the need for external speed sensors.

[0011] Preferably, the phase angular velocity parameter in the phase inertial deduction module is a state variable stored in the controller register. When the signal-to-noise ratio index is higher than the confidence threshold, the state variable is updated in real time based on the measured phase offset through the extended Kalman filter algorithm to characterize the motion trend of the cleaning machine relative to the periodic structural component.

[0012] Preferably, the spatial frequency analysis module, deviation compensation module, phase inertial deduction module, and slip suppression module in the system are integrated into the same industrial microcontroller; the industrial microcontroller is used to acquire horizontal ranging data sequences through a single single-line lidar and complete the calculation of phase offset and the generation of negative feedback torque commands within a millisecond-level control cycle.

[0013] Preferably, the deviation compensation module is also used to calculate the lateral distance deviation and angular deviation of the cleaning machine relative to the bottom surface of the periodic structural component based on the horizontal ranging data sequence, and input the lateral distance deviation into the first PID controller to generate a steering compensation command, and input the angular deviation into the second PID controller to generate a heading correction command. The motion compensation command is generated by superimposing the steering compensation command and the heading correction command.

[0014] Preferably, the system also includes an anomaly blocking module, which is used to monitor the duration of integral accumulation in real time. When the duration exceeds the preset safe time threshold and the signal-to-noise ratio index has not recovered to above the confidence threshold, it is determined that the environmental structural features are completely lost and the cleaning machine is forced to perform an in-situ braking operation to prevent the cumulative error from diverging.

[0015] Compared with the prior art, the beneficial effects of the present invention are: 1. This invention constructs a deterministic phase synchronization control mechanism based on environmental periodic constraints to solve the problem of cumulative divergence in long-distance operations of the cleaning machine. It takes the periodic physical structure in the confined space as the absolute reference system, transforms the probabilistic position estimation into one-dimensional phase synchronization control, extracts the fundamental phase from the environmental characteristic frequency domain analysis, and forces the longitudinal motion of the cleaning machine to be locked to the inherent period of the physical structure. It uses environmental geometric constraints to directly calibrate the state variables of the control system, so that the motion error of the cleaning machine is limited to a single structural period and does not accumulate with time or mileage. This enables long-distance precise operation without the support of a high-precision global map and eliminates the risk of control command jumps caused by feature similarity confusion. 2. This invention establishes a heterogeneous velocity field differential observation and torque closed-loop suppression system to maintain the motion stiffness of surfaces with low adhesion coefficients. To address the problem of odometer failure caused by wet and slippery conditions at the bottom of the tank, the phase differential is used to obtain the true feed speed relative to the environmental structure. This speed is used as the outer loop observation value and compared with the inner loop observation value based on wheel speed feedback. The differential signal is directly coupled to the bottom drive loop. When a slippage trend is detected, a negative feedback torque command is generated to modulate the output of the drive wheel in real time. The cross-level feedback control logic transforms environmental texture information into dynamic control constraints, enabling the cleaning machine's moving platform to perceive and suppress nonlinear idling of the actuator in real time, ensuring the stability of the control system's trajectory tracking in harsh fluid media environments. 3. A signal-to-noise ratio-gated virtual phase flywheel extrapolation strategy is adopted to improve the system's timing continuity and anti-interference capability in the blind zone. To address the risk of input signal interruption caused by bulkhead structural damage or obstruction, the control system incorporates a logic flywheel with state holding and inertial extrapolation functions. When the environmental characteristic signal-to-noise ratio is lower than the threshold, the system automatically switches to inertial extrapolation mode, using historical phase angular velocity to maintain the continuous operation of the control clock. After signal recovery, a smooth weighting algorithm is used to achieve seamless re-capture of the state, ensuring that the internal state estimation is consistent and the control command output is stable when external observation conditions change abruptly. This avoids the loss of discrete features that could cause the system to stop suddenly or the logic to reset, and improves the adaptability of the automated cleaning machine to non-standard structural environments. Attached Figure Description

[0016] Figure 1 This is a closed-loop logic flowchart of the autonomous navigation control system of the present invention; Figure 2 This is a comparison chart of trajectory tracking performance under different control strategies of the present invention; Figure 3 This is a diagram showing the hardware architecture and physical information interaction topology of the system of this invention. Detailed Implementation

[0017] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments. It should be noted that the description of these embodiments is for the purpose of helping to understand the present invention, but does not constitute a limitation thereof.

[0018] This invention provides an autonomous navigation and control system for a confined space cleaning machine, capable of handling cleaning operations in confined spaces under complex conditions such as repetitive internal structural features and slippery surfaces. The system includes a single-line lidar for collecting environmental data and a servo drive unit for executing motion control. It transforms the longitudinal position control of the cleaning machine within the confined space into tracking the fundamental phase of the periodic structural environment, and achieves slip suppression by monitoring the mapping deviation between the phase change rate and the rotational speed of the traveling mechanism. The system includes a spatial frequency analysis module for processing environmental perception data, which acquires information about the cleaning machine's operation. The system uses a sequence of horizontal ranging data from the side and employs a Fast Fourier Transform (FFT) algorithm to convert the time-domain ranging data into a frequency-domain signal. The module searches the spectrum for a specific frequency band with the highest amplitude, identifying this band as the fundamental frequency of the environmental periodic structural components, i.e., the bulkhead stiffeners or ribs. The spatial wavelength corresponding to this fundamental frequency characterizes the physical distance between adjacent structural components. The system utilizes a phase-locked loop (PLL) principle to calculate the phase difference between the peak signal in the current ranging data and the coordinate system of the cleaning machine, defining this phase difference as the phase offset. This process quantifies the longitudinal positional relationship of the cleaning machine relative to the environment. to The normalized phase value that changes periodically within the interval.

[0019] The system includes a phase inertial extrapolation module to maintain the continuity of the control clock. This module incorporates a state estimator based on an extended Kalman filter (EKF) and monitors the signal-to-noise ratio (SNR) of the fundamental frequency characteristics in real time. When the SNR exceeds a preset threshold, the system enters a locked mode, using the measured phase offset to iteratively update the phase angular velocity parameters stored in the controller register in real time. When the SNR falls below a confidence threshold, such as in situations involving high-concentration water mist or signal obstruction leading to feature loss, the system disables the output of the spatial frequency analysis module and switches to inertial extrapolation mode. In this mode, the module uses the most recently updated phase angular velocity parameters and the real-time speed of the cleaning machine to perform integral accumulation of the phase offset. The evolution of the internal control clock of the system is maintained by a virtual phase flywheel. The module also integrates smooth transition logic. When the signal-to-noise ratio index recovers to above the confidence threshold, the deviation between the deduced phase and the recaptured measured phase is calculated, and the deviation is eliminated within a preset transition period by a weighted average algorithm. In addition, the module has an abnormal blocking function. If the integral accumulation time exceeds the preset safety time threshold and the signal is not recovered, the cleaning machine is triggered to brake in place.

[0020] The system includes a slip suppression module to address slippage issues on surfaces with low adhesion coefficients. This module establishes a velocity observation mechanism based on phase differentiation, performing a time-dependent differential operation on the phase offset to obtain the phase change rate. And combined with the preset spacing parameters of the periodic structural components According to the operation rules Calculate the structure feed rate ,in Using pi as a constant, the structural feed speed characterizes the true linear velocity of the cleaning machine relative to the physical environment boundary. The module performs differential calculation between the structural feed speed and the theoretical linear velocity calculated based on the encoder feedback of the cleaning machine's traveling mechanism to obtain the observed slip velocity value. When the observed slip velocity value exceeds a preset slip threshold, the module generates a negative feedback torque command and superimposes it onto the drive control loop of the traveling mechanism. By reducing the torque output of the slipping drive wheel or adjusting the torque difference between the left and right drive wheels, the theoretical linear velocity converges to the structural feed speed. The system includes a deviation compensation module for trajectory tracking and attitude adjustment. This module identifies the boundary of the periodic structural component based on the fundamental wave characteristics and calculates the cleaning machine's return within the current structural cycle based on the phase offset. The module uses horizontal ranging data sequence to calculate the lateral distance and angular deviation of the cleaning aircraft relative to the bottom surface of the structural component. It employs a dual-closed-loop PID control architecture, inputting the lateral distance deviation into the first PID controller to generate steering compensation commands, and the angular deviation into the second PID controller to generate heading correction commands. These two are superimposed to generate motion compensation commands for the chassis. The module integrates a gain scheduling unit that dynamically adjusts the control weights based on the normalized progress value: when the normalized progress value indicates that the cleaning aircraft is approaching the boundary of the structural component, i.e., a peak or trough, the weight of the angular deviation control is increased to prevent collisions; when the normalized progress value indicates that the cleaning aircraft is located in the flat area in the middle of the structural component, the weight of the lateral distance deviation control is increased to optimize straight-line tracking performance.

[0021] Example 1: This example describes the application of an autonomous navigation control system for a tank cleaning machine in a confined space during bottom cleaning operations on a large crude oil tanker. This scenario features a corridor effect, with the working surface covered by a high-viscosity oil sludge mixture, accompanied by irregular high-concentration water mist interference generated by high-pressure cleaning jets. Under these conditions, the tank cleaning machine needs to perform long-distance linear propulsion and residue removal along the longitudinal reinforcing ribs at the bottom of the tank. The system utilizes a single-line lidar mounted on the side of the fuselage to acquire environmental information and relies on an industrial microcontroller to perform real-time calculations and control. For the geometric constraints formed by the periodically distributed reinforcing ribs within the cargo hold, the spatial frequency analysis module acquires a horizontal ranging data sequence and performs a fast Fourier transform. The system identifies the fundamental frequency representing the period of the reinforcing rib arrangement in the frequency domain. This frequency directly corresponds to the preset spacing parameters of the structural components in the physical environment. The system locks onto the fundamental frequency and maps the longitudinal physical displacement of the cleaning machine to a real-time value. to The phase offset within the interval cycle transforms absolute position measurement in physical space into relative synchronous tracking within the phase domain. This allows the system to maintain the monotonicity and continuity of longitudinal positioning by utilizing the periodic characteristics of the environment, even in the absence of a high-precision global map, thus avoiding the divergence of odometer cumulative errors caused by long-distance operations. Addressing the low adhesion coefficient and nonlinear slippage of the drive wheels due to sludge on the bilge surface, the slippage suppression module performs heterogeneous velocity field observation based on phase differentiation. When the cleaning machine's drive wheels idle due to sludge lubrication, although the theoretical linear velocity based on the motor encoder remains constant or increases, the actual displacement velocity of the cleaning machine relative to the reinforcing rib structure decreases. At this time, the slippage suppression module performs a time-dependent differential operation on the real-time calculated phase offset to obtain the phase change rate. And according to the operation rules Calculated structural feed rate ,in, Using pi as a constant, the system performs a differential comparison between the structure's feed speed and the theoretical linear velocity to obtain the observed slip velocity. Once the observed slip velocity exceeds the preset slip threshold, the system determines that slippage has occurred and generates a negative feedback torque command superimposed on the drive control loop. This negative feedback mechanism reduces the output torque of the drive wheel on the slipping side until the observed slip velocity converges to the allowable range. Thus, without the need for external contact speed measuring wheels or Doppler radar, the system achieves real perception of the body's motion state and traction control through phase analysis of the environmental structure texture. In response to the intermittent attenuation or obstruction of the lidar signal caused by high-pressure water mist during operation, the phase inertial inference module maintains the effectiveness of the control clock by monitoring the signal-to-noise ratio index of the fundamental wave characteristics. When the water mist concentration increases and the signal-to-noise ratio index falls below the preset threshold, the phase data output by the spatial frequency analysis module may jump or be lost. At this time, the system automatically shields the unstable real-time phase observation value and switches to the inertial inference mode.

[0022] In this mode, the system reads the phase angular velocity parameters updated and locked by the extended Kalman filter algorithm just before the signal-to-noise ratio drops. Combined with the current command speed of the cleaning machine, it continues to perform integral accumulation calculations on the phase offset. This virtual phase flywheel mechanism ensures that during the window of temporary blindness in external environmental perception, the system's internal control clock continues to evolve according to the predetermined physical inertia logic, maintaining continuous output of downstream control commands. When the water mist dissipates and the signal-to-noise ratio recovers to above the confidence threshold, the system smoothly transitions the inertial-derived phase to the recaptured measured phase within a preset transition period using a weighted averaging algorithm, achieving seamless switching between the perception blind zone and the observation zone. Regarding trajectory correction control, the deviation compensation module utilizes phase information... The system dynamically schedules the lateral control gain, identifies the boundaries of periodic structural components (i.e., the positions of peaks and troughs), and calculates the normalized progress value of the cleaning aircraft within the current structural cycle. When the normalized progress value indicates that the cleaning aircraft is running in the peak area of ​​a prominent structural component, the risk of collision increases due to the narrow physical space. The gain scheduling unit automatically increases the weight of angle deviation control, allowing the cleaning aircraft to prioritize maintaining a parallel course. When the normalized progress value indicates that the cleaning aircraft is running in the flat trough area between structural components, the system increases the weight of lateral distance deviation control, using a larger corrective torque to quickly eliminate lateral errors. This variable gain strategy based on the phase of the environmental structure enables the cleaning aircraft to adaptively adjust its control stiffness according to the tightness of local environmental constraints.

[0023] Example 2: This example aims to verify the engineering effectiveness of the autonomous navigation control system by constructing a simulated bilge environment containing a typical oil sludge mixture and high-pressure water mist interference. The test platform was built in a 20-meter-long and 3-meter-wide steel water tank. Reinforcing ribs with a height of 100 millimeters were welded to the bottom of the tank every 500 millimeters to physically reproduce the periodic structural characteristics of a standard cargo hold. A mixture of industrial grease and clay with a thickness between 20 and 50 millimeters was uniformly laid on the surface of the tank bottom to simulate the low-adhesion-coefficient oil sludge condition. During the test, high-pressure atomizing nozzles arranged above the water tank generated a water mist band with a visibility of less than 2 meters to introduce visual and optical ranging interference. The test design adopted a multi-dimensional control system, setting up a control group and the prototype of this invention. The control group was configured with a navigation system equipped with a traditional wheeled odometer and a two-dimensional lidar SLAM algorithm, without enabling the phase synchronization and slip suppression functions of this invention. The prototype of this invention fully utilized the spatial frequency analysis, phase inertial deduction, and slip suppression modules. Both systems were mounted on the same type of bilge cleaning machine chassis and started from the same initial position. The test was conducted under operating conditions, with a target linear propulsion speed of 0.2 m / s. The lateral deviation, longitudinal positioning error, and drive wheel speed fluctuations were recorded and compared in real time. The results showed that when the cleaning machine entered the sludge-covered area, the control group's drive wheel speed experienced severe high-frequency fluctuations, indicating slippage between the wheels and the working surface. Due to the lack of a targeted slippage suppression mechanism, the longitudinal positioning error of the control group exhibited a non-linear and rapid divergence trend with the travel distance. Data showed that by the time the machine reached 10 meters, the cumulative longitudinal error of the control group had exceeded 1.5 meters. Lateral control instability caused by positioning drift resulted in the fuselage repeatedly contacting the reinforcing rib sidewall. In contrast, the present invention's prototype, upon detecting a deviation between the phase change rate and the encoder feedback speed, rapidly intervened with the slippage suppression module. By actively reducing the torque output of the slipping drive wheel, the slippage rate was effectively controlled within 5%. Throughout the complete 20-meter journey, the longitudinal positioning error of the present invention's prototype was consistently limited to the distance between a single reinforcing rib (500 mm), and the maximum lateral deviation did not exceed 100 mm.

[0024] To verify the performance of the phase inertial inference module under extreme observation conditions, a high-concentration water mist was artificially introduced for 3 seconds to block the LiDAR point cloud data, resulting in a loss rate of over 90%. During this interference, the control group experienced a positioning jump due to the loss of environmental features, triggering the system's emergency stop protection logic. However, when the signal-to-noise ratio index of this invention dropped below the confidence threshold, it immediately and seamlessly switched to the inertial inference mode, using historical phase angular velocity to maintain the continuous operation of the control clock. After the water mist dissipated, the system quickly re-locked onto the environmental fundamental wave characteristics without any obvious control oscillations or path deviations. The velocity tracking accuracy under slip conditions and the control continuity index under water mist interference of this invention were both superior to the control group, strongly demonstrating the technical advantages of the heterogeneous velocity field closed-loop and phase inertial inference mechanism in solving the problems of low adhesion coefficient and weak observation (see Table 1).

[0025]

[0026] The above experimental results show that the present invention avoids the failure problem of traditional navigation schemes in confined space and harsh working conditions by transforming environmental geometric constraints into phase control variables and introducing a slip observation and inertial inference mechanism based on physical texture.

[0027] Example 3: This example combines Figures 1 to 3 The autonomous navigation and control system of a cabin cleaning machine in a confined space is described, such as... Figure 1 As shown, the logic flow of this autonomous navigation and control system begins with the raw ranging data acquired by a single-line lidar. This data, as a horizontal ranging data sequence, is input to the spatial frequency analysis module. This module performs operations such as extracting fundamental wave characteristics, identifying periodic structural boundaries, and calculating phase offset. The calculated fundamental wave characteristics and phase offset are transmitted to the phase inertial deduction module. This module is responsible for monitoring the signal-to-noise ratio (SNR) and activating the virtual phase flywheel to maintain the clock. Simultaneously, it updates the state using an extended Kalman filter algorithm and outputs the deduced phase change rate to the slip suppression module. It outputs the continuous phase and normalized progress to the deviation compensation module. The slip suppression module combines the theoretical linear velocity fed back from the walking mechanism and calculates the structural feed velocity using the phase change rate. After differential comparison of heterogeneous velocity fields, it generates a negative feedback torque command. The deviation compensation module combines the boundary identification data to calculate the lateral distance and angular deviation, performs gain scheduling based on the normalized progress, and generates motion compensation commands that include heading and steering corrections. Finally, the servo drive unit receives the motion compensation command and the negative feedback torque command, superimposes the commands, and controls the walking mechanism to perform physical motion, thereby completing the closed-loop control.

[0028] like Figure 2As shown, in a coordinate system with a longitudinal position of 0 to 20 meters and a lateral deviation of -0.5 to 0.5 meters, the trajectory tracking effects under different control strategies are demonstrated. The target trajectory is represented as a straight dotted line on the zero deviation axis, while the control group trajectory is represented as a large-span dashed line with fluctuations. This trajectory begins to separate from the zero axis at approximately 2 meters in the longitudinal position. As the longitudinal distance increases, the lateral deviation rapidly expands, reaching a peak deviation of approximately 0.45 meters in the 9 to 10 meter longitudinal position range. It shows a slow decreasing trend but still maintains a large error. In contrast, the trajectory of the sample group of this invention is represented as a solid line closely following the zero axis, remaining near the target trajectory throughout the entire 20-meter journey. The maximum lateral deviation fluctuates around 0.08 meters, and it exhibits convergence characteristics in the latter half of the journey after 10 meters in the longitudinal position. Figure 3 As shown, the system's hardware architecture mainly consists of four parts: an environmental sensing node, a core computing node, a power execution node, and a confined space operating environment. The environmental sensing node includes a single-line lidar and an inertial measurement unit (IMU), which are responsible for collecting environmental horizontal ranging data sequences and monitoring the aircraft's attitude and acceleration variance, respectively. The ranging sequence and attitude data are transmitted to the core computing node via a high-speed data interface. The core computing node is an industrial microcontroller with a real-time embedded operating system. Internally, it integrates spatial frequency analysis components, phase inertial inference components, deviation compensation components, and slip suppression components through an internal control bus, corresponding to fundamental frequency feature extraction, virtual clock maintenance, error correction, and slip control functions, respectively. The core computing node sends control signals containing motion compensation and negative feedback torque to the power execution node. The servo drive unit in the power execution node executes torque commands and speed closed loops accordingly, driving the walking mechanism, which includes an encoder, to move within the confined space operating environment. The walking mechanism then transmits the state feedback generated by the physical motion, namely encoder pulses and current, back to the computing node. At the same time, the periodic structural components and low-adhesion coefficient surfaces in the confined space operating environment react to the sensing node and the execution node through laser scanning and physical contact, respectively.

[0029] Example 4: This example provides an enhanced control method that includes adaptive parameter updates and boundary smoothing. It further refines the internal operating logic of the phase inertia derivation and slip suppression modules to ensure parameter determinism and control continuity across all operating conditions. This example specifies the phase angular velocity parameters. The update strategy is implemented when the signal-to-noise ratio (SNR) is higher than the preset threshold. In locked mode, phase angular velocity parameters Instead of relying on the current instantaneous phase difference, it iteratively estimates the phase shift using the state equation of a second-order extended Kalman filter (EKF). and phase angular velocity as a state vector And introduce the process noise covariance matrix. Covariance matrix of observation noise ,matrix The value of the matrix is ​​preset and calibrated based on the maximum acceleration and deceleration capability of the cleaning machine. It then dynamically adjusts in a negative correlation with the real-time signal-to-noise ratio (SNR) metric. When a decrease in the SNR metric is detected, but it remains above the threshold... At that time, the system automatically increases This reduces the weight of the current observation on state updates, thereby suppressing the impact of observation noise on the phase angular velocity parameter. Pollution.

[0030] To address the validity determination of slip velocity observations in the slip suppression module, this embodiment introduces a confidence interval evaluation mechanism based on the drive wheel dynamics model. The system pre-constructs a library of traction force-slip ratio characteristic curves for the cleaning machine on surfaces with different adhesion coefficients. During real-time operation, the slip suppression module not only calculates the slip velocity observations... Simultaneously calculate the current drive motor output torque and theoretical linear velocity. The ratio of the slip velocity to the equivalent traction drag coefficient is used to determine the current slip velocity observation value as having high confidence when the coefficient is in the linear stable region of the characteristic curve library, and it is directly used to generate the negative feedback torque command. When the coefficient enters the nonlinear saturation region or a step change occurs, the system determines that the slip velocity observation value may be affected by terrain changes, and a low-pass filter is introduced to smooth the negative feedback torque command. The cutoff frequency is set to the bandwidth of the drive system. to To prevent high-frequency oscillations in the control loop, and to address the potential control command jumps at the boundaries of periodic structural components in the deviation compensation module, this embodiment incorporates smooth transition logic based on the Sigmoid function into the gain scheduling unit. The system no longer uses hard threshold switching for angle deviation control weights. Lateral distance deviation control weight Instead, it is based on the normalized progress value. (range of values) The dynamic weight adjustment factor is generated according to the following rules. : ,in, The normalized position center of the structural component boundary. To adjust the slope parameter of the transition region width, the final control gain is calculated as follows: ,in, For boundary control gain, To control the gain in the central region, this logic ensures that the control weight can change continuously and monotonically when the cleaning machine transitions from the middle region of the structural component to the boundary region, eliminating the impact on the actuator caused by sudden changes in gain. In addition, for the triggering logic of the abnormal blocking module, this embodiment supplements a comprehensive judgment strategy based on multi-source information fusion. In addition to monitoring the duration of integral accumulation, the system also simultaneously monitors the current change rate of the drive motor and the acceleration variance of the IMU. When the integral time exceeds the limit, and the motor current is detected to be in the no-load or stall characteristic range, or the IMU data indicates that the fuselage attitude has been stationary for a long time, the system confirms that the environmental characteristics have been completely lost and triggers the stationary braking.

[0031] Example 5: To construct the traction force and slip ratio characteristic curve library required for the slip suppression module, a standardized offline data filling procedure must be executed before formal operation. This procedure controls the cleaning machine to perform a linear motion test with step-increase torque on the surface to be worked on, simultaneously recording the output torque of the drive motor and the structural feed speed calculated based on phase differential. The system employs the least squares method to fit the collected discrete data points to generate a continuous function curve describing the mapping relationship between the traction coefficient and the slip ratio. Then, by calculating the rate of change of the first derivative of this curve, the critical boundary between the linear stable region and the nonlinear saturation region under a specific working surface is quantified. For the parameter initialization of the extended Kalman filter algorithm in the phase inertial derivation module, the system executes a dynamic-static combined on-site calibration logic during the startup phase. In a static state, high-frequency sampling is used to determine the initial phase of the environmental fundamental wave characteristics and verify the preset spacing parameters. The matching degree is used to control the cleaning machine to travel at a constant low speed for a preset calibration distance in areas where the signal-to-noise ratio is higher than the confidence threshold. During this period, the residual sequence of the measured phase offset and the predicted state value is used to analyze the process noise covariance matrix. and observation noise covariance matrix The initial value is iteratively corrected until the covariance of the state estimate converges to the preset stable range.

[0032] Example 6: This example constructs a standardized engineering calibration and online status monitoring procedure for the autonomous navigation and control system of a cabin cleaning aircraft before deployment. This procedure covers baseline calibration of the spatial frequency analysis module, adaptive parameter initialization of the slip suppression module, and dynamic threshold tuning of the anomaly blocking module. It also includes preset spacing parameters for periodic structural components. In the engineering calibration, the system executes a calibration procedure based on spectral statistical baselines. This procedure controls the cleaning machine to acquire a horizontal ranging data sequence covering at least fifty structural cycles during the initial phase of operation. The spatial frequency analysis module performs a fast Fourier transform on the sequence to extract the dominant frequency component, and calculates the measured average spacing based on this component. Spacing from theoretical design Alignment generates linear correction factor The slip suppression module calculates the structural feed rate. Follow the modified physical constraint relationship , To determine the rate of change of phase offset with respect to time, the calibration procedure directly maps physical environment manufacturing tolerances and coating thickness variations to control parameters, ensuring the physical authenticity of velocity observations. For the spatial frequency analysis module, a baseline calibration process is performed before formal operation, placing the cleaning machine within a test section with standard periodic structural characteristics. The spacing parameters of the structural components in this test section are... After high-precision physical measurement and calibration, the system, once started, acquires a horizontal ranging data sequence for a duration of no less than 60 seconds in a static state, performs a fast Fourier transform, and extracts the fundamental frequency. The system calculates the measurement spacing parameters. With calibration parameters The deviation between them generates a frequency correction factor. The correction factor is written into the controller's non-volatile memory to perform linear compensation on all frequency domain analysis results during subsequent real-time operation, thereby avoiding systematic deviations caused by lidar installation errors or ranging nonlinearities.

[0033] For the slip suppression module, an online parameter adaptive initialization logic was established. In the initial stage when the cleaning machine enters the new working area, the system performs a low-speed uniform linear motion test with a speed set at 0.1 m / s. During this period, the system synchronously records the drive motor current. Compared with the theoretical linear velocity based on the encoder By calculating and storing the traction-velocity ratio sequence, the system constructs a basic adhesion coefficient model for the current working surface. This model is used to dynamically set the validity threshold for slip velocity observations, ensuring that the slip suppression function can intervene promptly when a slight slippage trend is detected on surfaces with different friction characteristics. Simultaneously, it avoids false triggers caused by road surface micro-textures. The confidence threshold in the phase inertial deduction module... Determined to adopt a method based on site environmental characteristics The statistical definition method involves continuously recording a fundamental frequency characteristic signal-to-noise ratio (SNR) data sequence under baseline conditions free from water mist interference before the cleaning machine begins formal operations, and then calculating the arithmetic mean of the dataset. with standard deviation Confidence threshold Set as This system triggers an inertial deduction mode when the signal quality deviates from the normal distribution statistical range, i.e., when the environmental signal-to-noise ratio (SNR) deteriorates substantially. This preserves tolerance for high-frequency noise in environmental textures while achieving a deterministic response to events involving the loss of structural features. For the abnormal blocking module, a dynamic threshold tuning strategy is implemented. The system monitors the historical distribution characteristics of the environmental SNR index in real time and calculates the mean. with standard deviation Based on the 3σ principle, the system dynamically adjusts the time threshold for abnormal blocking. ,in, Basic safety time, To adjust the coefficients and ensure that the system can appropriately relax its tolerance for short-term signal fluctuations and reduce unnecessary emergency stop operations under harsh operating conditions with a low overall signal-to-noise ratio, while tightening the threshold under operating conditions with a good signal-to-noise ratio to improve the response speed to sudden anomalies, the system can automatically adapt to different hardware individual differences and changes in environmental conditions through the execution of the above standardized procedures.

[0034] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0035] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. An autonomous navigation control system for a clean-out machine in a confined space, characterized by, The method comprises the following steps: a spatial frequency analysis module is configured to obtain a horizontal distance data sequence of a side of the ship unloader, and perform spatial frequency analysis on the horizontal distance data sequence to extract a fundamental wave feature and a current phase offset of a periodic structure within a limited space; a deviation compensation module is configured to identify a boundary of the periodic structure based on the fundamental wave feature, calculate a normalized progress value of the ship unloader within a current structure period in combination with the phase offset, and generate a motion compensation instruction for a chassis of the ship unloader based on the horizontal distance data sequence; a phase inertia deduction module is configured to calculate a signal-to-noise ratio index of the fundamental wave feature in real time, update a phase angular velocity parameter based on the continuously collected phase offset when the signal-to-noise ratio index is higher than a preset confidence threshold, and shield the output of the spatial frequency analysis module when the signal-to-noise ratio index is lower than the preset confidence threshold, and perform integral accumulation on the phase offset by using the phase angular velocity parameter and a real-time running speed of the ship unloader to maintain the continuity of a control clock; 2. The autonomous navigation control system for a cleanout machine in a confined space according to claim 1, wherein, a slip suppression module is configured to perform a differential operation on the phase offset with respect to time, and generate a structure feed speed in combination with a preset pitch parameter of the periodic structure; 3. The autonomous navigation control system for a cleanout machine in a confined space of claim 1, wherein, the structure feed speed is differentially calculated with a theoretical linear speed of a walking mechanism of the ship unloader to obtain a slip speed observation value, and a negative feedback torque instruction is generated and superimposed on a drive control loop of the walking mechanism of the ship unloader when the slip speed observation value exceeds a preset slip threshold to reduce the deviation between the theoretical linear speed and the structure feed speed.

4. The autonomous navigation control system for a cleanout machine in a confined space of claim 1, wherein, The spatial frequency analysis module is specifically configured to convert the time-domain horizontal distance data sequence into a frequency-domain signal, identify a specific frequency band with the highest amplitude as a fundamental wave frequency, confirm a data point corresponding to the fundamental wave frequency as the position of the periodic structure, and calculate a phase difference between a wave crest signal in the horizontal distance data sequence and a current position of the ship unloader as the phase offset.

5. The autonomous navigation control system for a cleanout machine in a confined space of claim 1, wherein, The slip suppression module follows the following operation rule when generating the structure feed speed: wherein, is defined as the structure feed speed, is defined as the differential rate of change of the phase offset with respect to time, is defined as the preset pitch parameter of the periodic structure, is the constant of the circle ratio; the slip suppression module maps the rate of change of the phase offset to the absolute linear speed of the cleaner relative to the physical boundary of the restricted space using the operation rule.

6. The autonomous navigation control system for a cleanout machine in a confined space of claim 1, wherein, The phase inertia deduction module is further configured to perform a smooth transition logic, calculate a phase deviation value between a deduction phase obtained by integral accumulation and a re-extracted measured phase when the signal-to-noise ratio index recovers from being lower than the confidence threshold to being higher than the confidence threshold, and gradually reduce the phase deviation value by a weighted average algorithm within a preset transition period until the system state is completely switched to a locking mode based on the measured feature. The deviation compensation module further comprises a gain scheduling unit configured to calculate the normalized progress value of the ship unloader within the current structure period based on the phase offset, and increase the weight of the angular deviation control when the normalized progress value indicates that the ship unloader is close to the boundary of the periodic structure, and increase the weight of the lateral distance deviation control when the normalized progress value indicates that the ship unloader is located in the middle region of the periodic structure. The slip suppression module is further configured to perform a torque redistribution operation, bypass a regular speed control instruction when the slip speed observation value exceeds the slip threshold, and directly reduce the torque output of the driving wheel where the slip occurs or adjust the torque difference between the left and right driving wheels until the slip speed observation value converges to within the slip threshold.

7. The autonomous navigation control system for a cleanout machine in a confined space of claim 1, wherein, The phase angular velocity parameter in the phase inertia deduction module is a state variable stored in a controller register, which is updated in real time by an extended Kalman filtering algorithm based on the measured phase offset when the signal-to-noise ratio index is higher than the confidence threshold, and represents the movement trend of the stripping machine relative to the periodic structure.

8. The autonomous navigation control system for a cleanout machine in a confined space of claim 1, wherein, The space frequency analysis module, the deviation compensation module, the phase inertia deduction module and the slip suppression module in the system are integrated in the same industrial microcontroller; the industrial microcontroller is used to collect a horizontal ranging data sequence through a single single-line laser radar, and to complete the calculation of the phase offset and the generation of the negative feedback torque instruction within a millisecond-level control cycle.

9. The autonomous navigation control system for a cleanout machine in a confined space of claim 1, wherein, The deviation compensation module is also used to calculate the lateral distance deviation and the angle deviation of the stripping machine relative to the bottom surface of the periodic structure based on the horizontal ranging data sequence, to input the lateral distance deviation into a first PID controller to generate a turning compensation instruction, and to input the angle deviation into a second PID controller to generate a heading correction instruction, and the motion compensation instruction is generated by superimposing the turning compensation instruction and the heading correction instruction.

10. The autonomous navigation control system for a cleanout machine in a confined space of claim 1, wherein, The system further comprises an abnormality blocking module, which is used to monitor the duration of the integral accumulation in real time, and when the duration exceeds a preset safety time threshold and the signal-to-noise ratio index has not recovered to above the confidence threshold, it is determined that the environmental structure feature is completely lost and the stripping machine is forced to perform a stationary braking operation to prevent the accumulation of errors from diverging.

Images