A wheelchair dynamic obstacle avoidance path planning method and system

Abstract

The present application relates to the technical field of wheelchair dynamic obstacle avoidance path planning, and more specifically, to a wheelchair dynamic obstacle avoidance path planning method and system. The method comprises: collecting pedestrian raw perception information, extracting the pedestrian's corresponding mobility limitation representation from the pedestrian raw perception information, determining the corresponding preset mobility limitation level based on the mobility limitation representation, wherein the mobility limitation representation reflects the presence of characteristic indicators that affect the pedestrian's normal walking; based on the mobility limitation level, calling the corresponding differentiated response strategy from a preset response strategy library, wherein the differentiated response strategy includes an avoidance action sequence and multimodal communication instructions; and continuously monitoring the pedestrian's raw perception information. When it is determined that the pedestrian has left the wheelchair's preset route, the differentiated response strategy is stopped. The method has the advantages of being able to provide differentiated avoidance and communication strategies, thereby improving the safety and efficiency of wheelchair interaction with such pedestrians in complex environments.

Description

Technical Field

[0001] The present application relates to the technical field of wheelchair dynamic obstacle avoidance path planning, and more specifically, to a wheelchair dynamic obstacle avoidance path planning method and system. Background Art

[0002] In modern urban traffic environments, wheelchairs are an essential means of transportation for people with limited mobility. Safe passage through complex scenarios, such as intersections without traffic lights, is a growing concern. To improve the intelligent capabilities of wheelchairs in such situations, existing technologies rely on perception systems that integrate multiple sensors, including vision, lidar, and ultrasound, to identify obstacles and pedestrians in real time. These systems, combined with algorithms, predict their movement trends to assist with path planning and dynamic obstacle avoidance.

[0003] However, existing wheelchair dynamic obstacle avoidance methods primarily rely on predicting pedestrians' typical traffic intentions (such as simple crossing or waiting patterns) and implementing standardized obstacle avoidance strategies that prioritize the wheelchair's own traffic efficiency. In real-world applications, pedestrian behavior at intersections is complex and diverse, and some individuals may exhibit atypical movement characteristics, placing higher demands on traffic efficiency and safety. Existing wheelchair obstacle avoidance systems may lack adaptability when handling such non-standard behaviors, making it difficult to fully assess traffic risks and make decisions, which in turn affects the interaction between wheelchairs and pedestrians and the overall traffic experience.

[0004] In view of the above problems, the existing technology is in urgent need of improvement. Summary of the Invention

[0005] The purpose of this application is to provide a wheelchair dynamic obstacle avoidance path planning method and system, which has the advantage of being able to provide differentiated avoidance and communication strategies, thereby improving the safety and efficiency of wheelchair interaction with such pedestrians in complex environments.

[0006] In a first aspect, the present application provides a method for dynamic obstacle avoidance path planning for a wheelchair, the method comprising:

[0007] Collecting original pedestrian perception information, extracting the pedestrian's corresponding limited mobility representation from the original pedestrian perception information, and determining the corresponding preset limited mobility level based on the limited mobility representation. The limited mobility representation reflects the presence of characteristic indicators that affect the pedestrian's normal walking;

[0008] Based on the level of mobility limitation, the corresponding differentiated response strategy is called from the preset response strategy library. The differentiated response strategy includes avoidance action sequences and multimodal communication instructions.

[0009] Continuously monitor the pedestrian's original perception information, and stop executing the differentiated response strategy when it is determined that the pedestrian has left the preset route of the wheelchair.

[0010] Through the above solution, differentiated avoidance and communication strategies can be provided for pedestrians with limited mobility, improving the safety and efficiency of wheelchair interaction with such pedestrians in complex environments.

[0011] Furthermore, the present application also proposes that the multimodal communication instructions include parameters for controlling projection brightness in the ground projection mode and parameters for controlling volume of voice prompts;

[0012] The method also includes:

[0013] Collect ambient light intensity and ambient noise intensity;

[0014] Adjust the projection brightness parameters according to the ambient light intensity;

[0015] Adjust the volume parameters according to the intensity of the ambient noise.

[0016] Through the above solution, the effectiveness of multimodal communication is improved.

[0017] Furthermore, the present application also proposes that the method further comprises:

[0018] Collecting instantaneous ground conditions, including local slopes, slippery areas, and / or uneven areas;

[0019] Based on the immediate ground conditions, the execution risk of the avoidance action sequence under the immediate ground conditions is evaluated according to the preset risk assessment rules;

[0020] Dynamically adjust the execution control parameters of the avoidance action sequence based on the execution risk.

[0021] Through the above solution, the safety of the avoidance action is improved.

[0022] Furthermore, the present application proposes collecting original pedestrian perception information, extracting the pedestrian's corresponding mobility restriction representation from the original pedestrian perception information, and determining the corresponding preset mobility restriction level based on the mobility restriction representation, including:

[0023] Extract multiple representations of limited mobility based on the original pedestrian perception information obtained during a preset observation period.

[0024] Analyze the frequency and / or duration of the same manifestations of limited mobility during the observation period;

[0025] According to the frequency and / or duration, the mobility limitation levels corresponding to the multiple mobility limitation representations within the observation time period are determined by a preset level calculation rule.

[0026] Through the above scheme, the accuracy of determining the level of restricted mobility is improved.

[0027] Furthermore, this application also proposes extracting the corresponding pedestrian's limited mobility representation from the pedestrian's original perception information, including:

[0028] Monitoring changes in environmental conditions, which are information that affects the quality of data collected from original pedestrian perception information;

[0029] Analyze change information based on preset environmental impact rules;

[0030] When the change information indicates that the data quality of the collected pedestrian original perception information has deteriorated, the pedestrian original perception information with deteriorated data quality is compensated according to the type of environmental conditions to generate adjusted pedestrian original perception information;

[0031] The corresponding limited mobility representation is extracted based on the adjusted original pedestrian perception information.

[0032] Through the above scheme, the accuracy of extracting the representation of limited mobility is improved.

[0033] Furthermore, the present application proposes to perform compensation processing on the original pedestrian perception information with degraded data quality according to the type of environmental conditions, including:

[0034] According to the types of various environmental conditions, parameterized compensation rules for compensation processing corresponding to the types of environmental conditions are preset;

[0035] According to the type of environmental condition, a quantitative indicator of the type of the preset environmental condition is obtained;

[0036] According to the type of environmental condition and the corresponding quantitative index, the compensation parameter is obtained through the parameterized compensation rule corresponding to the type of environmental condition;

[0037] The original pedestrian perception information with degraded data quality is compensated according to the compensation parameters.

[0038] Through the above solution, the precision and effectiveness of data compensation are improved.

[0039] Furthermore, the present application also proposes to determine the preset route for pedestrians to leave the wheelchair, including:

[0040] According to the level of mobility limitation, a target determination condition corresponding to the level of mobility limitation is determined from a plurality of preset groups of determination conditions, the plurality of groups of determination conditions including a spatial determination parameter for defining a traffic safety requirement and a time determination parameter for defining a stable state duration;

[0041] The relative position and motion state of the pedestrian are obtained from the pedestrian's original perception information. When the relative position and motion state are within the time defined by the time judgment parameter of the target judgment condition and meet the passage safety requirements defined by the space judgment parameter of the target judgment condition, the preset passage route of the pedestrian leaving the wheelchair is determined.

[0042] Through the above solution, the accuracy and adaptability of the judgment of the timing of the stop strategy are improved.

[0043] Furthermore, the present application proposes collecting original pedestrian perception information, extracting the pedestrian's corresponding mobility restriction representation from the original pedestrian perception information, and determining the corresponding preset mobility restriction level based on the mobility restriction representation, including:

[0044] Collecting original pedestrian perception information, determining that the original pedestrian perception information contains multiple pedestrians, and extracting the corresponding limited mobility representations of the multiple pedestrians;

[0045] Determining the mobility limitation level corresponding to each pedestrian based on the mobility limitation representations corresponding to the multiple pedestrians;

[0046] The mobility limitation level corresponding to each pedestrian is calculated comprehensively according to the preset comprehensive calculation rules to obtain the comprehensive mobility limitation level;

[0047] According to the level of action capability limitation, calling the corresponding differentiated response strategy from the preset response strategy library includes: according to the level of comprehensive action capability limitation, calling the corresponding differentiated response strategy from the preset response strategy library.

[0048] Through the above scheme, the comprehensiveness and applicability of the obstacle avoidance strategy in multi-pedestrian scenarios are improved.

[0049] Furthermore, this application also proposes collecting original pedestrian perception information and extracting the corresponding pedestrian's limited mobility representation from the original pedestrian perception information, including:

[0050] Collecting original perception information of pedestrians;

[0051] Based on the pedestrian's original perception information, the spatial relationship between the potential mobility-impeded representation and the pedestrian is identified;

[0052] Identify the motion relationship between the potential mobility-impeded representation and the pedestrian subject based on the pedestrian's original perception information;

[0053] Determine whether the object representing the restricted mobility is a representation of the pedestrian's restricted mobility based on the spatial relationship and movement relationship.

[0054] Through the above scheme, the accuracy and reliability of the recognition of the characterization of limited mobility are improved.

[0055] In a second aspect, the present application also proposes a wheelchair dynamic obstacle avoidance path planning system, comprising:

[0056] An acquisition and evaluation module is used to collect raw pedestrian perception information, extract the pedestrian's corresponding mobility restriction representation from the raw pedestrian perception information, and determine the corresponding preset mobility restriction level based on the mobility restriction representation. The mobility restriction representation reflects the presence of characteristic indicators that affect the pedestrian's normal walking;

[0057] A strategy calling module is used to call the corresponding differentiated response strategy from the preset response strategy library according to the level of action capability limitation. The differentiated response strategy includes avoidance action sequences and multimodal communication instructions;

[0058] The monitoring and strategy management module is used to continuously monitor the pedestrian's original perception information and stop executing the differentiated response strategy when it is determined that the pedestrian has left the preset route of the wheelchair.

[0059] From the above, it can be seen that the present application provides a wheelchair dynamic obstacle avoidance path planning method and system, which solves the problem of insufficient adaptability to pedestrians with limited mobility in the existing technology by identifying the characteristics of pedestrians' limited mobility and calling differentiated response strategies according to their levels. It has the advantage of being able to provide differentiated avoidance and communication strategies for pedestrians with limited mobility, thereby improving the safety and efficiency of wheelchair interactions with such pedestrians in complex environments. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] Figure 1 A flowchart of a method for dynamic obstacle avoidance path planning for a wheelchair is provided in one embodiment of the present application.

[0061] Figure 2 This is one of the flow charts of a wheelchair dynamic obstacle avoidance path planning method provided in another embodiment of the present application.

[0062] Figure 3 The second flowchart of a method for dynamic obstacle avoidance path planning for a wheelchair is provided in another embodiment of the present application.

[0063] Figure 4 The third flowchart of a method for dynamic obstacle avoidance path planning for a wheelchair provided in another embodiment of the present application.

[0064] Figure 5 This is a fourth flow chart of a method for dynamic obstacle avoidance path planning for a wheelchair provided in another embodiment of the present application.

[0065] Figure 6 FIG5 is a flowchart of a method for dynamic obstacle avoidance path planning for a wheelchair provided in another embodiment of the present application.

[0066] Figure 7FIG6 is a flowchart of a method for dynamic obstacle avoidance path planning for a wheelchair provided in another embodiment of the present application.

[0067] Figure 8 FIG7 is a flowchart of a method for dynamic obstacle avoidance path planning for a wheelchair provided in another embodiment of the present application.

[0068] Figure 9 FIG8 is a flowchart of a method for dynamic obstacle avoidance path planning for a wheelchair provided in another embodiment of the present application.

[0069] Figure 10 A flowchart of a wheelchair dynamic obstacle avoidance path planning system provided in another embodiment of the present application. DETAILED DESCRIPTION

[0070] The technical solutions in this application will be clearly and completely described below in conjunction with the drawings in this application. Obviously, the described embodiments are only a part of the embodiments of this application, rather than all of the embodiments. The components of the present application generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of the present application provided in the drawings is not intended to limit the scope of the application for which protection is claimed, but merely represents selected embodiments of the present application. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without making creative work are within the scope of protection of this application.

[0071] It should be noted that similar reference numerals and letters represent similar items in the following drawings. Therefore, once an item is defined in one drawing, it does not need to be further defined or explained in subsequent drawings. At the same time, in the description of this application, the terms "first", "second", etc. are only used to distinguish the description and should not be understood as indicating or implying relative importance.

[0072] When traditional wheelchair dynamic obstacle avoidance methods perceive pedestrians who exhibit clear and observable limitations on their mobility due to reasons such as carrying heavy objects, using assistive devices, pushing strollers, or supporting others, they are not adaptable enough to such special pedestrians because they rely mainly on conventional pedestrian models. As a result, they are unable to effectively assess their special safety needs, fail to provide targeted and differentiated avoidance strategies, and fail to clearly communicate their own intentions. This may cause traffic conflicts, unnecessary traffic delays, or reduce the wheelchair user's sense of security and trust.

[0073] As an example during the development of this application, consider a wheelchair equipped with an advanced sensor system approaching an unsignaled intersection at a steady speed, intending to pass through it in a straight line. Its sensors detect a pedestrian in the pedestrian waiting area ahead of the intersection. The pedestrian faces the wheelchair's path and shows signs of starting to move. The wheelchair's pedestrian interaction intention recognition module initially determines that the pedestrian has a high probability of crossing and marks it as a potential collision target. The central control system then initiates standard obstacle avoidance procedures, such as slightly slowing down and calculating alternative detour trajectories. However, during continued observation, the wheelchair's sensor system may detect atypical features of the pedestrian. For example, through detailed image analysis, the system may notice that the pedestrian is carrying seemingly heavy shopping bags in both hands, limiting upper limb swing. Or, gait analysis may reveal that the pedestrian's steps are slower and more labored than average, or even swaying when starting, indicating that the weight is affecting their balance and agility. Other possible scenarios include the pedestrian using a cane, supporting an elderly person with more limited mobility, or pushing a stroller with a young child. These characteristics, derived from sensor data analysis and reflecting the actual mobility of pedestrians, are collectively referred to as "pedestrian mobility limitation characteristics." These characteristics directly indicate that the pedestrian's physical mobility, reaction speed, and flexibility in changing direction may be significantly lower than those of healthy, unburdened adults. In such intersection scenarios, when the wheelchair's pedestrian interaction intention recognition system perceives that the pedestrian has the above-mentioned obvious mobility limitation characteristics and determines that the pedestrian's mobility is significantly lower than that of ordinary pedestrians due to their own objective reasons, the existing wheelchair dynamic obstacle avoidance method, which mainly relies on predicting and standardizing the pedestrian's regular passage intentions and taking the wheelchair's own passage efficiency as an important consideration, exposes its serious lack of adaptability to the special behavior patterns of such pedestrians with clear "mobility limitation characteristics."

[0074] Reference Figure 1 , this application proposes a wheelchair dynamic obstacle avoidance path planning method, including:

[0075] S100: Collecting original pedestrian perception information, extracting a corresponding pedestrian's mobility limitation representation from the original pedestrian perception information, and determining a corresponding preset mobility limitation level based on the mobility limitation representation, wherein the mobility limitation representation reflects characteristic indicators that affect the pedestrian's normal walking;

[0076] S800: Based on the level of mobility limitation, a corresponding differentiated response strategy is called from a preset response strategy library. The differentiated response strategy includes an avoidance action sequence and multimodal communication instructions.

[0077] S900: Continuously monitor the pedestrian's original perception information and stop executing the differentiated response strategy when it is determined that the pedestrian has left the preset route of the wheelchair.

[0078] Among them, pedestrian original perception information refers to the original data about surrounding pedestrians obtained through the sensor system carried by the wheelchair. It can be obtained by various perception devices such as visual sensors, lidar, ultrasonic sensors or millimeter wave radars. It is mainly used to obtain basic information about the pedestrian's position, movement status, appearance characteristics, etc.

[0079] The preset level of mobility restriction refers to a pre-set classification of the pedestrian's mobility based on the representation of mobility restriction. Different levels can be set according to different representations of mobility restriction or their combinations, such as mild, moderate, severe, etc. It is mainly used to quantitatively evaluate the pedestrian's mobility so as to subsequently call for differentiated response strategies.

[0080] The preset response strategy library refers to a database or storage structure that stores preset differentiated response strategies for different levels of mobility limitations. It can contain multiple sets of pre-designed avoidance action sequences and multimodal communication instructions. It is mainly used to quickly call corresponding avoidance and communication plans based on the assessed level of mobility limitations.

[0081] A differentiated response strategy refers to an avoidance and communication plan customized according to the level of mobility limitations of pedestrians. It includes avoidance action sequences and multimodal communication instructions. It is mainly used to provide more adaptable and safe avoidance interactions for pedestrians with different mobility capabilities.

[0082] An avoidance action sequence refers to a series of preset or calculated motion control instructions executed by a wheelchair to avoid conflicts with pedestrians. It may include a combination of deceleration, stopping, turning, detouring, etc. It is mainly used to adjust the movement trajectory and speed of the wheelchair to ensure a safe distance from pedestrians.

[0083] Multimodal communication instructions refer to instructions that convey wheelchair intentions or prompt information to pedestrians through multiple modes. They can include visual prompts (such as ground projections, screen displays), auditory prompts (such as voice broadcasts, prompt sounds), etc. They are mainly used to communicate explicitly with pedestrians and enhance the clarity and predictability of interactions.

[0084] The preset route refers to the predetermined driving path of the wheelchair before performing obstacle avoidance. It can be a straight line, a preset curve or a path planned by the navigation system. It is mainly used to define the normal driving range of the wheelchair and serve as a reference for judging whether the pedestrian has left the conflict area.

[0085] The core innovation of this application lies in collecting the original perception information of pedestrians and extracting the limited mobility representation from it, thereby determining the preset limited mobility level, and calling the differentiated response strategy in the preset response strategy library according to the level, thereby specifically solving the problem of insufficient adaptability of existing wheelchair obstacle avoidance methods to pedestrians with limited mobility, and achieving the effect of improving obstacle avoidance safety and interaction efficiency.

[0086] The specific operating principle is as follows: First, the wheelchair's sensor system continuously collects raw sensory information about pedestrians in the surrounding environment. This information includes various data such as the pedestrian's position, movement, and appearance. Based on this raw sensory information, the system further analyzes and extracts possible signs of mobility limitations, such as whether the pedestrian is carrying heavy objects, using assistive devices, or has an abnormal gait. Subsequently, based on these extracted signs of mobility limitations, the system compares them with pre-defined rules or models to determine the pedestrian's corresponding pre-defined mobility limitation level, thereby assessing and grading the pedestrian's actual mobility and potential risk. Based on the determined mobility limitation level, the system invokes a differentiated response strategy from a library of pre-defined response strategies. This strategy includes a set of avoidance maneuver sequences and multimodal communication commands tailored for that pedestrian type. The wheelchair then begins executing this differentiated response strategy, adjusting its own motion (executing the avoidance maneuver sequence) and sending communication signals to the pedestrian (executing the multimodal communication commands), enabling safer and clearer interaction with the pedestrian. During the implementation of the differentiated response strategy, the system continuously monitors the pedestrian's raw sensory information and tracks the pedestrian's position and status in real time. If the system determines that the pedestrian has left the wheelchair's preset route and is no longer in the potential conflict zone, the wheelchair ceases its current differentiated response strategy and resumes normal travel or makes a decision based on the new environmental information. This entire process forms a closed loop of perception, assessment, decision-making, execution, and re-monitoring, ensuring that the wheelchair can dynamically and intelligently adapt to pedestrians of varying mobility, improving the safety and efficiency of obstacle avoidance.

[0087] In one specific implementation of this embodiment, a wheelchair can be equipped with sensors such as a camera, lidar, and microphone to collect raw sensory information from the pedestrian. The system can analyze the camera data using image processing algorithms to identify the pedestrian's posture, belongings, or assistive devices, and extract indicators of limited mobility. For example, it can identify a cane, a stroller, or a noticeable gait instability. Based on the identified indicators, the system can consult a preset level comparison table and map these indicators to preset levels of limited mobility. For example, mapping "using a cane" to "moderately limited" and "pushing a stroller" to "mildly limited." The preset response strategy library can store multiple sets of avoidance trajectories and corresponding voice prompts or ground projection patterns. For example, for the "moderately limited" level, a preset detour trajectory with a greater safety distance can be invoked, and a voice prompt can be triggered to inform the pedestrian that the wheelchair is about to detour. The system continuously tracks the pedestrian's position using lidar. When the pedestrian's position exceeds the safety boundaries on both sides of the wheelchair's preset route, the system determines that the pedestrian has left and stops executing the current detour trajectory and voice prompt.

[0088] Reference Figure 2 ,Furthermore, the multimodal communication instructions include ,control projection brightness parameters for the ground projection mode and ,control volume parameters for the voice prompt;

[0089] The method also includes:

[0090] S200: collecting ambient light intensity and ambient noise intensity;

[0091] S300: Adjusting and controlling projection brightness parameters according to ambient light intensity;

[0092] S400: Adjust the volume parameters according to the ambient noise intensity.

[0093] Among them, the projection brightness control parameter refers to the numerical value or signal used to set the light intensity output by the ground projection device. It can be implemented as an integer value or percentage representing the brightness level, or a voltage / current signal that controls the power of the projection light source. Its purpose is to adjust the visibility of the ground projection on the ground. The volume control parameter refers to the numerical value or signal used to set the sound volume output by the voice prompt device. It can be implemented as an integer value or percentage representing the volume level, or an electrical signal that controls the gain of the audio amplifier. Its purpose is to adjust the audibility of the voice prompt. The ambient light intensity refers to the intensity of natural light or artificial lighting in the wheelchair's surrounding environment, which can be quantified in units such as lux or lumens. The ambient noise intensity refers to the size of the background sound in the wheelchair's surrounding environment, which can be quantified in units such as decibels. Acquisition refers to obtaining real-time values ​​of ambient light intensity and ambient noise intensity through sensors or data interfaces. Adjustment refers to calculating and setting new projection brightness control parameters and volume control parameters based on the collected environmental parameter values ​​and in accordance with preset rules or algorithms.

[0094] By collecting real-time light and noise levels from the wheelchair's environment and dynamically adjusting the ground projection brightness and voice prompt volume parameters in multimodal communication instructions based on this environmental data, the wheelchair can adaptively optimize the information transmission between it and pedestrians based on the specific environmental conditions. For example, in bright outdoor conditions, the system can automatically increase the brightness of the ground projection to ensure clear visibility; on noisy streets, the system can automatically increase the volume of voice prompts to ensure pedestrians can hear the prompts. Conversely, in dim or quiet conditions, the system can appropriately reduce the brightness and volume to avoid unnecessary light pollution or noise interference. This ability to dynamically adjust communication parameters based on environmental factors enables the multimodal communication component of the wheelchair to function more effectively when executing differentiated response strategies based on the pedestrian's level of mobility limitations. This ensures that the wheelchair's avoidance intentions and passage information are accurately and clearly conveyed to pedestrians, especially those with limited mobility, thereby enhancing the robustness and effectiveness of the entire obstacle avoidance method and reducing the potential risk of conflict caused by information transmission barriers.

[0095] In one specific implementation of this embodiment, the wheelchair is equipped with an ambient light sensor and a microphone. The ambient light sensor measures the ambient light intensity in front of the wheelchair in real time and outputs an electrical signal or digital value representing the light intensity. The microphone collects ambient sound in real time, calculates the ambient noise intensity through an audio processing unit, and outputs a numerical value representing the noise intensity. After receiving the light intensity and noise intensity values, the wheelchair control system can consult a preset lookup table or execute a mapping function. For example, the lookup table can map different light intensity ranges to different projection brightness levels (i.e., control projection brightness parameters) and different noise intensity ranges to different voice volume levels (i.e., control volume parameters). The control system then sends the calculated or searched control projection brightness parameters to the ground projection module to control it to adjust the projection brightness; and sends the control volume parameters to the voice prompt module to control it to adjust the voice playback volume.

[0096] Reference Figure 3 In other embodiments of the present application, the method further includes:

[0097] S500: collecting real-time ground conditions, where the real-time ground conditions include local slopes, slippery areas, and / or uneven areas;

[0098] S600: Evaluate the execution risk of the avoidance action sequence under the instantaneous ground conditions according to preset risk assessment rules;

[0099] S700: Dynamically adjust execution control parameters of the avoidance action sequence based on the execution risk.

[0100] Specifically, immediate ground conditions refer to the ground conditions of the area where the wheelchair is currently located or about to pass through, including local slopes, slippery areas, and / or uneven areas. Acquisition of immediate ground conditions can be achieved through a variety of sensors. For example, local slopes can be obtained through inertial measurement units or tilt sensors; slippery areas can be identified by analyzing ground reflectivity or texture features using humidity sensors or visual sensors; and uneven areas can be detected by ultrasonic sensors, lidar, or visual sensors to detect ground undulations or obstacles. Preset risk assessment rules refer to logic or models used to quantitatively assess the potential danger level when executing an avoidance action sequence under specific immediate ground conditions. Specifically, they can be a series of conditional judgment statements, table lookup rules, calculation formulas based on physical models, or machine learning models. These rules comprehensively consider the type and degree of ground conditions and the specific parameters of the avoidance action sequence to output a risk value or risk level. Execution risk refers to the assessed likelihood or severity of an accident when executing the avoidance action sequence under the immediate ground conditions. It can be a numerical value or a level. Its purpose is to provide a basis for subsequent parameter adjustments. Dynamically adjusting the execution control parameters of an avoidance maneuver sequence involves modifying or optimizing the kinematic or dynamic control parameters of the avoidance maneuver sequence in real time based on the assessed execution risk. Specifically, this involves adjusting the wheelchair's speed, acceleration, deceleration, turning angular velocity, steering angle, or selecting a different avoidance maneuver sequence. This aims to reduce the risk of executing an avoidance maneuver under current ground conditions and ensure the wheelchair completes the avoidance process safely and stably.

[0101] By collecting the wheelchair's current, real-time ground conditions, including local slopes, slippery areas, and / or uneven surfaces, the system obtains real-time environmental information that impacts the wheelchair's stability. Based on these real-time ground conditions, the system quantifies the potential execution risk of the avoidance maneuver sequence under these conditions according to pre-set risk assessment rules. For example, if the ground is slippery and a sharp turn is required, the risk assessment rules will determine a high risk. The system then dynamically adjusts the execution control parameters of the avoidance maneuver sequence based on the assessed execution risk. For example, if the risk assessment indicates a high risk, the system can reduce the wheelchair's speed, reduce the turning angle, or select a smoother avoidance path. In this way, this solution can perceive environmental changes in real time, assess the safety of avoidance maneuvers, and promptly adjust the wheelchair's motion control, significantly reducing the risk of slipping, tipping, or loss of control when executing avoidance maneuvers under complex ground conditions. Furthermore, by combining the basis for determining differentiated avoidance maneuver sequences based on the pedestrian's level of mobility impairment, this solution, after determining the avoidance maneuver sequence for a specific pedestrian, does not blindly execute it, but instead takes the wheelchair's specific driving environment into account. By perceiving and assessing the immediate ground conditions and risks, this solution ensures that the selected avoidance action sequence is safe and feasible in the current actual environment. For example, for a pedestrian with a higher level of mobility impairment, the system may select an avoidance action sequence with a larger reserved space and a slower speed. However, if the ground is found to be slippery and sloping when executing this sequence, this solution will further adjust the speed and turning parameters of the sequence so that it can be executed safely even on slippery slopes. This strategy, which combines the characteristics of the pedestrian and the wheelchair environment, allows the wheelchair's obstacle avoidance behavior to meet the needs of special pedestrians while ensuring its own safety in complex environments. It effectively solves the technical problem of safely implementing differentiated avoidance strategies for pedestrians with limited mobility under complex ground conditions.

[0102] In one specific implementation of this embodiment, the wheelchair is equipped with multiple sensors, such as a bottom-mounted inclination sensor for measuring local slope, a humidity sensor mounted near the drive wheel for detecting ground humidity, and a front-mounted visual sensor for analyzing ground texture and undulations to identify slippery and uneven areas. When the wheelchair needs to execute an evasive maneuver sequence, for example, a sequence involving deceleration and a left turn determined based on the perceived level of mobility impairment of a pedestrian, the system first collects current and immediate ground conditions. Suppose the inclination sensor detects a rightward slope, the humidity sensor detects high ground humidity, and the visual sensor analyzes the ground texture and determines it is slippery. Based on this information, the system performs an assessment according to a preset risk assessment rule. This rule can be a lookup table that determines the risk level based on the slope, humidity, and type of evasive maneuver. For example, the lookup table may indicate that the risk of executing a left turn under the current slope and humidity conditions is "medium risk." Based on this "medium risk" assessment, the system dynamically adjusts the execution control parameters of the evasive maneuver sequence. Specifically, the system can reduce the planned turning angular velocity and further reduce the wheelchair's speed to minimize the possibility of slipping or tipping when turning on slippery slopes. The adjusted parameters are sent to the wheelchair's motion controller to execute a modified avoidance maneuver sequence.

[0103] Reference Figure 4 In another embodiment of the present application, step S100 includes:

[0104] S110: extracting multiple mobility limitation representations based on the acquired pedestrian original perception information within a preset observation time period;

[0105] S120: Analyze the frequency and / or duration of the same manifestation of limited mobility during the observation period;

[0106] S130: Determine the mobility limitation levels corresponding to the plurality of mobility limitation representations within the observation time period according to the frequency and / or duration and a preset level calculation rule.

[0107] Among them, the preset observation time period refers to a certain length of time used to collect and analyze the original perception information of pedestrians, which can be implemented by a time window of fixed length; the frequency and / or duration refers to the number of times a specific limited mobility representation appears and the length of time the representation remains in existence within the preset observation time period, which can be obtained by a counter or timer; the preset level calculation rule refers to an established algorithm or logic for determining the level of limited mobility based on the frequency and / or duration of the limited mobility representation, which can be implemented by a threshold-based judgment rule, a weighted sum model or a machine learning model.

[0108] Over a preset observation period, the system continuously collects raw pedestrian sensory information and extracts multiple mobility restriction signatures. These signatures may appear at different times or persist for a period of time. The system then analyzes the same mobility restriction signatures extracted within the same observation period, counting their frequency and / or duration. For example, if the system detects a pedestrian's gait being unstable multiple times within a given observation period, or detects a pedestrian's walking speed below a certain threshold for a sustained period, the system then applies a preset ranking rule based on these statistically calculated frequency and / or duration data to comprehensively evaluate these dynamically changing signatures and determine a ranking that reflects the pedestrian's overall mobility restriction during the observation period. This approach, by conducting observations and statistical analysis over a period of time, avoids the instability associated with judgments based on instantaneous states. Frequency reflects the repetitiveness and significance of the signatures, while duration reflects their stability. This combined consideration allows for more accurate distinction between occasional abnormal behavior and persistent mobility restriction. This more accurate mobility restriction ranking method, based on time-period analysis, provides more reliable input for subsequent differentiated response strategy invocation and policy management. When the system can accurately determine the actual degree of a pedestrian's mobility limitations, it can call upon avoidance action sequences and multimodal communication instructions that better match that level from a library of preset response strategies. For example, for pedestrians with higher levels of mobility limitations, the system may invoke more conservative avoidance actions and more explicit communication instructions, thereby more effectively ensuring the safe passage of such pedestrians and avoiding unnecessary conflicts and delays. This improvement makes the entire wheelchair dynamic obstacle avoidance path planning method more adaptable and safe when dealing with pedestrians with limited mobility, enhancing the practicality and reliability of the overall solution.

[0109] In a specific implementation of this embodiment, the preset observation time period is set to 5 seconds. During these 5 seconds, the system continuously collects the original perception information of the pedestrian and extracts the characterization of limited mobility from it. For example, the system may extract characterizations such as unstable gait, slow walking speed, and body shaking at different times. The system then analyzes the characterizations extracted within these 5 seconds, and statistically shows that the unstable gait occurred 2 times, the slow walking speed lasted for 3 seconds, and the body shaking occurred 1 time. According to the preset level calculation rules, the rules can be set as follows: If the frequency of unstable gait is greater than 1 time or the duration of slow walking speed is greater than 2 seconds, the level of limited mobility is determined to be moderately limited. Based on the statistical results, the system determines that the level of limited mobility of the pedestrian is moderately limited.

[0110] Reference Figure 5 In step S100, extracting the pedestrian's corresponding limited mobility representation from the pedestrian's original perception information includes:

[0111] A1: Monitors changes in environmental conditions, which are information that affects the quality of data collected from original pedestrian perception information.

[0112] A2: Analyze change information based on preset environmental impact rules;

[0113] A3: When the change information indicates that the data quality of the collected pedestrian original perception information has degraded, the pedestrian original perception information with degraded data quality is compensated according to the type of environmental conditions to generate adjusted pedestrian original perception information;

[0114] A4: Extract the corresponding limited mobility representation based on the adjusted original pedestrian perception information.

[0115] Environmental condition change information refers to information about changes in environmental factors that affect the quality of sensor data collection, such as changes in light intensity, visibility, background noise, and ground reflections. Environmental conditions refer to external environmental factors that affect the data quality of collected pedestrian raw perception information, such as light, rain, fog, snow, wind, background noise, and ground conditions. Preset environmental impact rules refer to a set of pre-established rules that describe how different environmental conditions affect the quality of specific sensor data. These rules can be based on experimental data, empirical knowledge, or physical models. For example, they specify the degree to which the image signal-to-noise ratio decreases under a specific light intensity, or the range within which the lidar ranging error increases under a specific rainfall or fog concentration. Compensation processing refers to the process of correcting or enhancing pedestrian raw perception information whose data quality has been degraded due to environmental conditions using specific algorithms or technical means. This can be achieved using image enhancement algorithms, signal filtering algorithms, data fusion algorithms, or model-based correction methods. Adjusted pedestrian raw perception information refers to pedestrian raw perception information whose data quality has been improved after compensation processing. This information is closer to the data quality under ideal conditions and can be used for subsequent more accurate feature extraction.

[0116] By monitoring changing environmental conditions, the system promptly detects environmental changes that may affect data quality. Based on pre-set environmental impact rules, this information is analyzed to determine whether the environmental changes have caused data quality degradation. If data quality is determined to have degraded, the affected pedestrian's original perception information is compensated based on the type of environmental conditions to generate adjusted pedestrian perception information. This process aims to eliminate or mitigate environmental impacts on the data. Finally, the corresponding mobility limitation representation is extracted based on the adjusted pedestrian perception information. The improved input data quality also increases the accuracy of the extracted representation. This solution improves the mobility limitation representation extraction step in the basic method by implementing environmentally adaptive compensation, making the extracted representation more accurate and reliable. A more accurate representation more precisely reflects the pedestrian's actual mobility, enabling subsequent steps such as determining the mobility limitation level and invoking differentiated response strategies to be based on more reliable information. This in turn enables more appropriate avoidance maneuver sequences and multimodal communication commands, ultimately improving the effectiveness and safety of the entire wheelchair dynamic obstacle avoidance path planning method.

[0117] In one specific implementation of this embodiment, the wheelchair is equipped with a light sensor and a visibility sensor to monitor changes in environmental conditions. For example, when the light sensor reading falls below a certain threshold, or the visibility sensor indicates an increase in fog concentration, the system can identify a change in environmental conditions. Preset environmental impact rules can be stored in a rule base, including rules such as "When light intensity falls below X lux, the image signal-to-noise ratio decreases by Y dB" and "When visibility falls below Z meters, the lidar ranging error increases by W%." Based on the sensor readings and the rules, the system can determine whether the current light or visibility conditions have caused a decrease in visual or lidar data quality. When data quality is determined to have degraded, the system performs compensation based on the type of environmental conditions. For example, if light is insufficient, histogram equalization or gamma correction can be performed on the images captured by the visual sensor. If fog is present, distance-based filtering or a defogging algorithm can be applied to the lidar point cloud data. After processing, adjusted images or point cloud data are generated. Based on these adjusted data, the system can use an improved gait analysis algorithm to extract characteristics such as cadence and stride length, or identify pedestrian postures or auxiliary tools based on the processed point cloud.

[0118] Reference Figure 6 Furthermore, sub-step A32 of step A3: compensating the original pedestrian perception information with degraded data quality according to the type of environmental conditions, including:

[0119] A321: According to the types of various environmental conditions, parameterized compensation rules corresponding to the compensation processing of each environmental condition are preset;

[0120] A322: According to the type of environmental condition, obtain the quantitative index of the type of the preset environmental condition;

[0121] A323: Based on the type of environmental condition and the corresponding quantitative index, a compensation parameter is obtained by using a parameterized compensation rule corresponding to the type of environmental condition;

[0122] A324: Compensate the original pedestrian perception information with degraded data quality based on compensation parameters.

[0123] The parameterized compensation rule of compensation processing refers to a compensation algorithm or model designed for a specific type of environmental condition, which can adapt to different degrees of environmental influences by adjusting parameters. It can be implemented using mathematical functions, lookup tables or machine learning models; the quantitative index of the type of environmental condition refers to a numerical value used to measure the degree of influence of a specific type of environmental condition, such as light intensity (lux), rainfall (mm / hour), visibility (m), etc., which can be obtained using sensor measurements or the output of an environmental assessment algorithm; the compensation parameter refers to a numerical value or vector calculated by the parameterized compensation rule based on the type and quantitative index of the environmental condition, and used to specifically perform the compensation processing, such as gain, offset, filter coefficient in image processing, or noise suppression intensity in point cloud data, etc. Its purpose is to fine-tune the compensation process according to the specific conditions of the environment.

[0124] Compensation is performed on degraded raw pedestrian perception information based on the type of environmental condition. Specifically, first, parameterized compensation rules corresponding to each environmental condition are pre-set. This means that for each environmental condition that may affect the quality of raw pedestrian perception information, a specific set of compensation rules with adjustable parameters is pre-designed. Second, a quantitative indicator of the pre-set environmental condition type is obtained based on the environmental condition type. This is used to quantitatively assess the environmental condition and better apply the parameterized compensation rule. The quantitative indicator can be used to convert the impact of the environmental condition into a specific numerical value, providing a basis for subsequent compensation processing. Then, based on the environmental condition type and the corresponding quantitative indicator, compensation parameters are obtained using the parameterized compensation rule corresponding to the environmental condition type. This involves combining the environmental condition type and quantitative indicator with the pre-set parameterized compensation rule to calculate the specific compensation parameters. Different environmental condition types and quantitative indicators correspond to different compensation parameters, thus achieving differentiated compensation processing. Finally, compensation is performed on degraded raw pedestrian perception information based on the compensation parameters. This calculated compensation parameter is applied to the raw perception information to correct data deviations caused by the environmental condition. It is precisely because different parameterized compensation rules are used for different types of environmental conditions, and combined with quantitative indicators of environmental conditions, that refined compensation processing of pedestrians' original perception information is achieved. This improves the quality of the compensated data, thereby providing more reliable basic data for the subsequent extraction of corresponding limited mobility representations based on the adjusted pedestrians' original perception information.

[0125] In one specific implementation of this embodiment, when the environmental condition is identified as low light, the system can obtain a quantitative indicator of the current ambient light intensity, such as the lux value measured by a light sensor. The preset parameterized compensation rule for low light conditions can be a function that calculates gain and offset compensation parameters for the image data based on the lux value. For example, when the lux value is low, a larger gain and offset value is calculated; when the lux value is moderate, a smaller gain and offset value is calculated. The calculated gain and offset parameters are then applied to each pixel in the original image data to enhance brightness and adjust contrast, thereby generating adjusted image data that is clearer and more detailed in low-light environments. For example, when the environmental condition is identified as rainy, the system can obtain a quantitative indicator of rainfall, such as the rainfall rate estimated by a rain sensor or visual analysis. The preset parameterized compensation rule for rainy conditions can be an image deraining algorithm model that parameterizes the deraining intensity based on the rainfall rate. For example, a higher rainfall rate results in a greater deraining intensity. Then, the calculated deraining strength parameter is applied to the original image data to remove the influence of raindrops and rain streaks, generating cleaner adjusted image data.

[0126] Reference Figure 7 In step S900, determining the preset route for the pedestrian to leave the wheelchair includes:

[0127] B1: Based on the level of mobility limitation, determine the target determination condition corresponding to the level of mobility limitation from multiple sets of preset determination conditions. The multiple sets of determination conditions include spatial determination parameters for defining traffic safety requirements and temporal determination parameters for defining the duration of state stability.

[0128] B2: The relative position and motion state of the pedestrian are obtained from the pedestrian's original perception information. When the relative position and motion state are within the time defined by the time judgment parameter of the target judgment condition and meet the passage safety requirements defined by the space judgment parameter of the target judgment condition, the preset passage route for the pedestrian to leave the wheelchair is determined.

[0129] The preset multiple sets of judgment conditions refer to a series of rule sets that are pre-set for different levels of mobility limitations to determine whether pedestrians leave the preset routes. They can be stored in the wheelchair control system and organized in the form of tables, databases or rule sets; target judgment conditions refer to the set of judgment conditions selected from the preset multiple sets of judgment conditions for specific judgment based on the currently identified level of mobility limitations of pedestrians; spatial judgment parameters refer to parameters used to define the spatial range or position relationship of traffic safety requirements, which may include the minimum safe distance between pedestrians and wheelchairs, the allowable deviation range of the pedestrian's area relative to the preset route, whether the pedestrian enters or leaves a specific area, etc.; time judgment parameters refer to parameters used to define the length of time a state needs to continue to meet the spatial judgment requirements. It can be a fixed time interval or a time threshold that is dynamically adjusted according to other factors.

[0130] Based on the identified pedestrian's mobility impairment level, the system searches a pre-set judgment condition library for a target judgment condition that matches that level. This target judgment condition includes the spatial safety requirements set for pedestrians of that specific level and the duration for which these requirements must be met. The system then continuously acquires the pedestrian's real-time relative position and motion status information. This real-time information is then compared with the selected target judgment condition. Only when the pedestrian's relative position and motion status continuously meet the spatial safety requirements specified in the target judgment condition within the specified timeframe does the system ultimately determine that the pedestrian has left the wheelchair's preset route. This mechanism for determining whether a pedestrian has left the preset route, combined with the basis for invoking differentiated response strategies based on the pedestrian's mobility impairment level, provides a mechanism for accurately determining when to terminate the differentiated response strategy after execution. By using differentiated spatial and temporal judgment thresholds based on the pedestrian's specific mobility impairment level to determine whether the pedestrian has truly left the danger zone, the system avoids the potential for misjudgment or delayed response that can result from using a single standard. This differentiated judgment mechanism, combined with the execution of differentiated response strategies, enables wheelchairs to stop executing avoidance actions and communication instructions in a timely and safe manner and resume normal passage when pedestrians no longer pose a potential conflict. This improves the passage efficiency of wheelchairs while ensuring the safety of pedestrians with different mobility abilities.

[0131] In a specific implementation of this embodiment, the level of limited mobility can be divided into three levels: "mild", "moderate", and "severe". Different judgment conditions can be preset for these levels. For example, for the "mild" level, the spatial judgment parameter can be set to the distance between the pedestrian and the wheelchair is greater than 2 meters, and the angle between the pedestrian's movement direction and the wheelchair path is greater than 90 degrees; the time judgment parameter can be set to last for 1 second. For the "moderate" level, the spatial judgment parameter can be set to the distance between the pedestrian and the wheelchair is greater than 3 meters, and the angle between the pedestrian's movement direction and the wheelchair path is greater than 120 degrees; the time judgment parameter can be set to last for 2 seconds. For the "severe" level, the spatial judgment parameter can be set to the distance between the pedestrian and the wheelchair is greater than 4 meters, and the angle between the pedestrian's movement direction and the wheelchair path is greater than 150 degrees; the time judgment parameter can be set to last for 3 seconds. When the system recognizes that the level of limited mobility of the pedestrian is "moderate", it will select the corresponding judgment condition. Then the pedestrian's position and movement status will be continuously monitored. Only when the distance between the pedestrian and the wheelchair remains greater than 3 meters for 2 consecutive seconds and the angle between the pedestrian's movement direction and the wheelchair path remains greater than 120 degrees, will the system determine that the pedestrian has left the wheelchair's preset route and stop executing the differentiated response strategy for "moderate" level pedestrians.

[0132] Reference Figure 8 Furthermore, when there are multiple pedestrians on the road, step S100 includes:

[0133] C1: Collect original pedestrian perception information, determine whether the original pedestrian perception information contains multiple pedestrians, and extract the corresponding limited mobility representations of multiple pedestrians;

[0134] C2: Determine the mobility limitation level of each pedestrian based on the corresponding mobility limitation representations of multiple pedestrians;

[0135] C3: Calculate the mobility limitation level corresponding to each pedestrian according to the preset comprehensive calculation rules to obtain the comprehensive mobility limitation level;

[0136] The steps include calling the corresponding differentiated response strategy from the preset response strategy library based on the level of action capability limitation, including:

[0137] C4: Based on the level of comprehensive action capability limitation, the corresponding differentiated response strategy is called from the preset response strategy library.

[0138] Specifically, in this embodiment, the preset comprehensive calculation rule refers to a preset algorithm or logic for integrating the mobility restriction levels of multiple pedestrians to obtain an overall assessment result. This can be achieved through methods such as taking the highest level, weighted averaging, or comprehensive assessment based on a risk model. The comprehensive mobility restriction level refers to the overall assessment level obtained by comprehensively calculating the mobility restriction levels of multiple pedestrians according to the preset comprehensive calculation rule. This can be achieved through a single level or multi-dimensional indicators reflecting the overall risk of a group.

[0139] By collecting raw pedestrian perception information, the system first determines whether the information contains multiple pedestrians. The ability to identify multiple pedestrians allows subsequent processing to be tailored to the group rather than the individual. Once multiple pedestrians are identified, the system extracts the corresponding mobility impairment representations for each pedestrian and, based on these representations, independently determines each pedestrian's mobility impairment level. This separate assessment ensures accurate understanding of each individual's situation. Based on this, the solution further calculates the mobility impairment levels of each pedestrian into a single, comprehensive mobility impairment level according to pre-set comprehensive calculation rules. This comprehensive calculation comprehensively considers the overall travel risks and needs of multiple pedestrians, avoiding potential strategic errors caused by focusing solely on individual pedestrians. Finally, based on this comprehensive mobility impairment level, a corresponding, differentiated response strategy is invoked from a pre-set response strategy library. This comprehensive level-based strategy invocation allows the wheelchair to adopt obstacle avoidance maneuvers and communication methods that take into account the overall situation of the group when encountering multiple pedestrians. This allows for more effective handling of complex multi-pedestrian scenarios, improves the rationality and effectiveness of obstacle avoidance strategies, and reduces potential travel risks. Compared with the basic solution that only handles a single pedestrian, this solution adds the recognition of multi-pedestrian scenes, individual evaluation, group comprehensive evaluation, and strategy invocation based on group evaluation, which significantly enhances the wheelchair's obstacle avoidance ability in complex multi-pedestrian environments and can better ensure traffic safety and efficiency.

[0140] In one specific implementation of this embodiment, the wheelchair's visual sensors and lidar continuously collect environmental data, generating raw pedestrian perception information. The processing unit analyzes this information, using target detection and tracking algorithms to identify all pedestrians within its field of view and determine whether there are multiple pedestrians. For each identified pedestrian, the system further analyzes their image and motion characteristics, for example, using a gait analysis algorithm to identify abnormal gait and an image recognition algorithm to detect whether they are carrying large objects or using assistive devices, thereby extracting a representation of each pedestrian's mobility impairment. Based on preset rules or models, for example, pedestrians with a slow gait or using a cane are assessed as moderately impaired, while pedestrians pushing strollers or assisting elderly people are assessed as mildly impaired. The system then determines the mobility impairment level for each pedestrian based on preset comprehensive calculation rules, such as taking the highest mobility impairment level among all pedestrians as the overall mobility impairment level, or performing a weighted calculation based on the number of pedestrians, their location distribution, and their respective levels. For example, if there is one pedestrian with a moderate impairment and one pedestrian with a mild impairment, the overall level might be determined as moderately impaired. Finally, based on this comprehensive level of mobility limitation, the system searches and invokes a corresponding differentiated response strategy from a library of pre-set response strategies. For example, if the comprehensive level is moderately limited, the system might invoke a strategy that includes a longer avoidance distance, a lower avoidance speed, and multimodal communication instructions, including both voice prompts and ground projection warnings.

[0141] Reference Figure 9 In step S100, original pedestrian perception information is collected and the corresponding pedestrian's limited mobility representation is extracted from the original pedestrian perception information, including:

[0142] D1: Collecting original pedestrian perception information;

[0143] D2: Based on the pedestrian’s original perception information, identify the spatial relationship between the potential mobility-impeded representation and the pedestrian’s main body;

[0144] D3: Identify the motion relationship between the potential mobility-impeded representation and the pedestrian subject based on the pedestrian's original perception information;

[0145] D4: Determine whether the object representing the restricted mobility is a representation of the pedestrian's restricted mobility based on spatial and motion relationships.

[0146] Specifically, the spatial relationship refers to the geometric relationship between the potential mobility impairment and the pedestrian, including relative position, distance, and orientation. This can be described using parameters such as three-dimensional coordinates, relative distance, and angle. The motion relationship refers to the relative motion state between the potential mobility impairment and the pedestrian, such as relative speed, relative acceleration, and similarity or synchronization of motion trajectories. This can be described using motion vectors and trajectory matching.

[0147] By collecting raw pedestrian sensory information, raw data about the pedestrian is acquired. Then, based on this raw sensory information, the spatial relationship between the potential mobility impairment indicator and the pedestrian's main body is identified, and their spatial correlation is determined. Simultaneously, the kinematic relationship between the potential mobility impairment indicator and the pedestrian's main body is identified based on the raw sensory information, and their kinematic correlation is determined. Finally, by comprehensively analyzing the spatial and kinematic relationships, it is determined whether the potential mobility impairment indicator is truly associated with the pedestrian's movements, thereby determining whether it is a representation of the pedestrian's limited mobility. For example, if a potential mobility impairment indicator (such as a crutch) is spatially close to the pedestrian's main body and its movement is highly synchronized with that of the pedestrian's main body, the crutch can be determined to be a representation of the pedestrian's limited mobility. Conversely, if the crutch is near the pedestrian but has no close spatial or kinematic correlation with the pedestrian, it is not considered a representation of the pedestrian's limited mobility. By incorporating spatial and kinematic relationship analysis, this method improves the accuracy of identifying mobility impairment indicators from raw sensory information and can distinguish between limitations caused by the pedestrian themselves and those caused by environmental factors or unrelated objects. This makes it more reliable to subsequently determine the level based on the characterization of limited mobility and invoke differentiated response strategies, thereby more effectively dealing with pedestrians with special mobility patterns and helping to ensure the safety of the obstacle avoidance process.

[0148] In a specific implementation of this embodiment, the wheelchair's visual sensor and lidar can collect images and point cloud data of pedestrians and strollers. Image processing and target recognition algorithms can identify the pedestrian body and the stroller as potential representations of restricted mobility. The system can calculate the distance and relative position between the stroller and the pedestrian body, for example, the stroller is in front of the pedestrian and the distance is less than a certain threshold, to determine the spatial relationship. The system can track the movement trajectories of the pedestrian and the stroller, calculate their relative speed and direction of movement, for example, the stroller and the pedestrian body are moving in the same direction and the relative speed is close to zero, to determine the motion relationship. Based on the close spatial relationship and synchronous motion relationship between the stroller and the pedestrian, the system can determine that the stroller is a representation of the pedestrian's restricted mobility.

[0149] Reference Figure 10 , this application further proposes a wheelchair dynamic obstacle avoidance path planning system, including:

[0150] An acquisition and evaluation module is used to collect raw pedestrian perception information, extract the pedestrian's corresponding mobility restriction representation from the raw pedestrian perception information, and determine the corresponding preset mobility restriction level based on the mobility restriction representation. The mobility restriction representation reflects the presence of characteristic indicators that affect the pedestrian's normal walking;

[0151] A strategy calling module is used to call the corresponding differentiated response strategy from the preset response strategy library according to the level of action capability limitation. The differentiated response strategy includes avoidance action sequences and multimodal communication instructions;

[0152] The monitoring and strategy management module is used to continuously monitor the pedestrian's original perception information and stop executing the differentiated response strategy when it is determined that the pedestrian has left the preset route of the wheelchair.

[0153] Among them, the acquisition and evaluation module refers to the functional unit responsible for acquiring external environmental information and performing preliminary processing. It can be implemented by integrating multiple sensor interfaces, data preprocessing circuits and embedded processors; the strategy calling module refers to the functional unit that selects and outputs corresponding execution instructions based on input information. It can be implemented by a memory that stores a preset strategy library and a decision-making unit that performs table lookup or logical judgment based on the input; the monitoring and strategy management module refers to the functional unit that continuously tracks the target state and controls according to preset conditions. It can be implemented by a sensor data interface, a state judgment logic circuit and a control signal output interface.

[0154] The solution of this application implements the wheelchair dynamic obstacle avoidance path planning method in a systematic manner by concretizing each step into functional modules. The acquisition and evaluation module is responsible for acquiring the pedestrian's raw sensory information and analyzing the pedestrian's mobility limitations, providing basic data for subsequent obstacle avoidance strategy selection. The strategy invocation module invokes the corresponding differentiated response strategy from a preset response strategy library based on the mobility limitation level determined by the acquisition and evaluation module. The monitoring and strategy management module continuously monitors the pedestrian's status and promptly stops executing the current differentiated response strategy when the pedestrian leaves the wheelchair's preset route. Through the collaborative operation of these three modules, information flows between modules. The processing results of the acquisition and evaluation module are transmitted to the strategy invocation module for decision-making, and the monitoring and strategy management module dynamically manages the execution of the strategy based on real-time monitoring results. This modular system design enables the effective implementation of the wheelchair dynamic obstacle avoidance path planning method. Through functional division, it improves the system's reliability and efficiency, thus solving problems that are difficult to achieve with a single method.

[0155] The above description is merely an embodiment of the present application and is not intended to limit the scope of protection of the present application. For those skilled in the art, various modifications and variations of the present application are possible. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the scope of protection of the present application.

Claims

1. A wheelchair dynamic obstacle avoidance path planning method, characterized in that: The method comprises: Collecting raw pedestrian perception information, extracting a corresponding pedestrian's mobility limitation representation from the raw pedestrian perception information, and determining a corresponding preset mobility limitation level based on the mobility limitation representation, wherein the mobility limitation representation reflects the presence of characteristic indicators that affect the pedestrian's normal walking; Based on the level of mobility limitation, a corresponding differentiated response strategy is called from a preset response strategy library. The differentiated response strategy includes an avoidance action sequence and a multimodal communication instruction. The multimodal communication instruction includes a control parameter for projection brightness of a ground projection mode and a control parameter for voice prompt volume. Collect ambient light intensity and ambient noise intensity; Adjusting the control projection brightness parameter according to the ambient light intensity; Adjusting the control volume parameter according to the intensity of the ambient noise; The original perception information of the pedestrian is continuously monitored, and when it is determined that the pedestrian has left the preset passage route of the wheelchair, the differentiated response strategy is stopped.

2. The wheelchair dynamic obstacle avoidance path planning method according to claim 1, characterized in that: The method further comprises: collecting instant ground conditions, wherein the instant ground conditions include local slopes, slippery areas, and / or uneven areas; According to the instantaneous ground conditions, evaluating the execution risk of the avoidance action sequence under the instantaneous ground conditions according to a preset risk assessment rule; Dynamically adjust execution control parameters of the avoidance action sequence according to the execution risk.

3. The wheelchair dynamic obstacle avoidance path planning method according to claim 1, characterized in that: The collecting of original pedestrian perception information, extracting a corresponding pedestrian's mobility limitation representation from the original pedestrian perception information, and determining a corresponding preset mobility limitation level based on the mobility limitation representation include: Extracting a plurality of said limited mobility representations based on the obtained original perception information of the pedestrian within a preset observation time period; Analyzing the frequency and / or duration of occurrence of the same manifestation of limited mobility within the observation time period; The mobility limitation levels corresponding to the plurality of mobility limitation representations within the observation time period are determined according to the frequency and / or duration and by a preset level calculation rule.

4. The wheelchair dynamic obstacle avoidance path planning method according to claim 1, characterized in that: The extracting of the pedestrian's corresponding limited mobility representation from the pedestrian's original perception information includes: Monitoring information on changes in environmental conditions, wherein the environmental conditions are environmental information that affects the quality of data collected from the original pedestrian perception information; Analyze the change information according to preset environmental impact rules; When the change information indicates that the data quality of the collected pedestrian original perception information has deteriorated, compensating the pedestrian original perception information with the deteriorated data quality according to the type of the environmental condition to generate adjusted pedestrian original perception information; A corresponding limited mobility representation is extracted based on the adjusted original perception information of the pedestrian.

5. The wheelchair dynamic obstacle avoidance path planning method according to claim 4, characterized in that: The compensating process for the original pedestrian perception information with degraded data quality according to the type of the environmental condition includes: According to the types of the various environmental conditions, parameterized compensation rules for compensation processing corresponding to the types of the various environmental conditions are preset; According to the type of the environmental condition, obtaining a preset quantitative index of the type of the environmental condition; According to the type of the environmental condition and the corresponding quantitative index, a compensation parameter is obtained by using the parameterized compensation rule corresponding to the type of the environmental condition; Compensation processing is performed on the original pedestrian perception information with degraded data quality according to the compensation parameters.

6. The wheelchair dynamic obstacle avoidance path planning method according to claim 1, characterized in that: The method of determining the preset route for the pedestrian to leave the wheelchair includes: Determining, based on the mobility limitation level, a target determination condition corresponding to the mobility limitation level from a plurality of preset determination conditions, the plurality of determination conditions comprising a spatial determination parameter for defining a traffic safety requirement and a temporal determination parameter for defining a state stability duration; The relative position and motion state of the pedestrian are obtained from the pedestrian's original perception information. When the relative position and motion state meet the passage safety requirements defined by the spatial judgment parameters of the target judgment conditions within the time period defined by the time judgment parameters of the target judgment conditions, the preset passage route of the pedestrian leaving the wheelchair is determined.

7. The wheelchair dynamic obstacle avoidance path planning method according to claim 1, characterized in that: The collecting of original pedestrian perception information, extracting a corresponding pedestrian's mobility limitation representation from the original pedestrian perception information, and determining a corresponding preset mobility limitation level based on the mobility limitation representation include: collecting the original pedestrian perception information, determining that the original pedestrian perception information includes multiple pedestrians, and extracting the limited mobility representations corresponding to the multiple pedestrians; determining the mobility limitation level corresponding to each pedestrian based on the mobility limitation representations corresponding to the plurality of pedestrians; Calculate the mobility limitation level corresponding to each pedestrian according to a preset comprehensive calculation rule to obtain a comprehensive mobility limitation level; The calling of a corresponding differentiated response strategy from a preset response strategy library according to the action capability limitation level includes: calling a corresponding differentiated response strategy from a preset response strategy library according to the comprehensive action capability limitation level.

8. The wheelchair dynamic obstacle avoidance path planning method according to claim 1, characterized in that: The collecting of original pedestrian perception information and extracting the corresponding pedestrian's limited mobility representation from the original pedestrian perception information includes: Collecting original perception information of the pedestrian; identifying a spatial relationship between a representation of a potential impeded mobility and a pedestrian subject based on the original perception information of the pedestrian; identifying a motion relationship between a representation of a potential impeded mobility and a pedestrian subject based on the pedestrian's original perception information; It is determined whether the mobility-impeded representation object is a representation of the pedestrian's limited mobility according to the spatial relationship and the motion relationship.

9. A wheelchair dynamic obstacle avoidance path planning system, characterized in that: include: a collection and evaluation module for collecting raw pedestrian perception information, extracting a corresponding pedestrian's mobility limitation representation from the raw pedestrian perception information, and determining a corresponding preset mobility limitation level based on the mobility limitation representation, wherein the mobility limitation representation reflects the presence of characteristic indicators that affect the pedestrian's normal walking; The acquisition and evaluation module is further used to acquire the ambient light intensity and ambient noise intensity; a strategy calling module, configured to call a corresponding differentiated response strategy from a preset response strategy library according to the level of mobility limitation, wherein the differentiated response strategy includes an avoidance action sequence and a multimodal communication instruction; The multimodal communication instructions include parameters for controlling projection brightness in a ground projection mode and parameters for controlling volume of voice prompts; The strategy calling module is further used to adjust the control projection brightness parameter according to the ambient light intensity; and is also used to adjust the control volume parameter according to the ambient noise intensity; The monitoring and strategy management module is used to continuously monitor the pedestrian's original perception information and stop executing the differentiated response strategy when it is determined that the pedestrian has left the preset travel route of the wheelchair.

Images