Remote control method and device for autonomous vending vehicles

Abstract

This invention relates to the field of intelligent vehicle control technology, and more particularly to a remote control method and device for autonomous vending vehicles. The invention automatically identifies events such as stockouts or the need for remote takeover through real-time sensing data, triggering intelligent decision-making or remote driving intervention on the business platform. Dual-link communication separates key control commands from massive amounts of sensing data, ensuring reliable transmission of remote control commands while providing the platform with real-time environmental awareness capabilities. This method transforms the traditional passive and inefficient operation mode, reliant on manual inspection and on-site handling, into a highly efficient management mode of proactive discovery, intelligent decision-making, and remote safe collaboration. This significantly improves vehicle operating efficiency and resource utilization, reduces reliance on manpower and maintenance costs, and enhances overall intelligence, service reliability, and customer satisfaction in complex scenarios.

Description

Technical Field

[0001] This invention relates to the field of intelligent vehicle control technology, and in particular to a remote control method and device for an autonomous vending vehicle. Background Technology

[0002] Currently, the operation and management of autonomous vending vehicles in scenic areas and other similar settings primarily rely on manual labor, leading to inefficiencies and slow response times. Specifically, inventory status can only be assessed through regular on-site inspections by operators or by estimating based on rough sales records, making real-time and accurate monitoring impossible. When vehicles stall due to sold-out items or obstacles, staff typically need to travel to the site to resolve the issue, resulting in high labor costs and a long timeframe from problem discovery to resolution. This significantly reduces the effective operating time of the vehicles, directly impacting sales revenue and service experience.

[0003] In summary, existing technological solutions are essentially a human-centric, passive response model, suffering from systemic defects such as poor real-time performance, high reliance on human labor, long operational downtime, and a lack of remote intervention capabilities. Therefore, there is an urgent need for an intelligent management method for autonomous vending vehicles to fundamentally improve operational efficiency, reduce labor costs, and enhance service reliability. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a remote control method and device for autonomous vending vehicles, so as to solve the above-mentioned technical problem.

[0005] The technical solution of this invention to solve the above-mentioned technical problems is as follows: A remote control method for an autonomous vending vehicle, comprising: during the operation of the autonomous vending vehicle, acquiring in real time the perception data corresponding to the autonomous vending vehicle, the perception data including data of the vehicle's surrounding environment, data of the status of goods in the cargo compartment, and data of the vehicle's operating status; wherein, the autonomous vending vehicle and the remote business platform are connected by a dual-link communication connection, the first link of the dual-link communication is used to transmit autonomous driving commands and status data, and the second link is used to transmit perception data and control commands related to remote driving; and performing event recognition based on the perception data to determine whether... A preset event is triggered, which includes at least a stockout event and a remote takeover event. When a stockout event is triggered, stockout event information is generated and reported to the remote business platform corresponding to the autonomous vending vehicle. Subsequently, a replenishment decision instruction issued by the remote business platform in response to the stockout event information is received and executed to replenish the autonomous vending vehicle. When a remote takeover event is triggered, a remote takeover request is generated and initiated to the remote business platform. Subsequently, a remote driving instruction issued by the remote business platform in response to the remote takeover request is received and executed to remotely control the autonomous vending vehicle.

[0006] The beneficial effects of this invention are as follows: This method automatically identifies stockouts or events requiring remote takeover through real-time sensing data, triggering the business platform to make intelligent decisions or remote driving interventions. Dual-link communication separates key control commands from massive amounts of sensing data, ensuring reliable transmission of remote control commands while providing the platform with real-time environmental awareness capabilities. This method transforms the traditional passive and inefficient operation mode, which relies on manual inspection and on-site handling, into a highly efficient management mode of proactive discovery, intelligent decision-making, and remote safe collaboration. This significantly improves vehicle operating efficiency and resource utilization, reduces reliance on manpower and maintenance costs, and enhances the overall intelligence level, service reliability, and customer satisfaction in complex scenarios.

[0007] Based on the above technical solution, the present invention can be further improved as follows.

[0008] Furthermore, the step of identifying events based on the perceived data to determine whether a preset event is triggered includes: extracting pressure time-series data corresponding to each product in the autonomous vending vehicle from the perceived data; for each product, inputting the pressure time-series data corresponding to the product into a trained time-series classification model to identify the product's state category through the time-series classification model; and determining that the out-of-stock event is triggered when any product maintains the out-of-stock state for multiple consecutive determination periods.

[0009] Furthermore, the step of identifying events based on the perceived data to determine whether a preset event is triggered further includes: determining, based on the vehicle operating status data, whether the speed of the autonomous vending vehicle is continuously lower than a speed threshold within a first preset time period; determining, based on the vehicle's surrounding environment data, whether the width of the target passage area of ​​the autonomous vending vehicle is continuously less than a width threshold within a second preset time period; performing a fusion decision to generate a takeover request confidence level when the speed is continuously lower than the speed threshold and the width of the target passage area is continuously less than the width threshold; and triggering the remote takeover event when the takeover request confidence level is greater than the confidence threshold.

[0010] Furthermore, the step of performing a fusion decision to generate a takeover request confidence score includes: generating a first sub-confidence score based on the duration of the vehicle speed being lower than a speed threshold and the speed difference between the vehicle speed and the speed threshold; generating a second sub-confidence score based on the duration of the target passage area width being less than a width threshold and the width difference between the target passage area width and the width threshold; and weighting and fusing the first sub-confidence score and the second sub-confidence score based on preset speed condition weights and width condition weights to obtain the takeover request confidence score.

[0011] Furthermore, the replenishment decision instruction is obtained in the following manner: based on the stockout event information, the comprehensive utility value corresponding to each preset candidate decision scheme is calculated; from all candidate decision schemes that meet the preset hard constraints, the candidate decision scheme with the highest comprehensive utility value is selected as the target decision scheme; based on the target decision scheme, the replenishment decision instruction is generated.

[0012] Further, the step of calculating the comprehensive utility value corresponding to each preset candidate decision scheme based on the stockout event information includes: calculating the expected revenue, operating cost, and execution risk corresponding to each preset candidate decision scheme based on the stockout event information; wherein, the expected revenue corresponding to each preset candidate decision scheme represents the expected revenue obtained by executing the preset candidate decision scheme, the operating cost corresponding to each preset candidate decision scheme represents the resource cost required to execute the preset candidate decision scheme, and the execution risk corresponding to each preset candidate decision scheme represents the risk of decreased customer satisfaction or task priority conflict caused by executing the preset candidate decision scheme; identifying the operating scenario of the autonomous vending vehicle based on the stockout event information; adjusting the preset weights according to the operating scenario to obtain new weights; and performing a weighted fusion calculation on each preset candidate decision scheme based on the expected revenue, operating cost, execution risk, and new weights to obtain the comprehensive utility value corresponding to the preset candidate decision scheme.

[0013] Furthermore, the preset weights include expected revenue weight, operating cost weight, and execution risk weight; adjusting the preset weights according to the operating scenario to obtain new weights includes: increasing the value of the expected revenue weight when the operating scenario is that the autonomous vending vehicle is in the target operating period; increasing the value of the execution risk weight when the operating scenario is that the autonomous vending vehicle is in the target operating area; and increasing the value of the operating cost weight when the operating scenario is that the battery power of the autonomous vending vehicle is lower than a preset threshold.

[0014] Furthermore, the remote driving command is obtained through the following methods: acquiring a scene snapshot data packet for the autonomous vending vehicle; generating a digital twin operating interface for the autonomous vending vehicle and its surrounding environment based on the scene snapshot data packet; calculating and presenting at least one alternative escape route in the digital twin operating interface using a path recommendation algorithm; and generating the remote driving command based on the user's interactive operations on the target location or the alternative escape route in the digital twin operating interface.

[0015] Furthermore, receiving and executing the remote driving command issued by the remote business platform in response to the remote takeover request includes: receiving the remote driving command issued by the remote business platform in response to the remote takeover request; performing a safety verification on the remote driving command based on the local perception data of the autonomous vending vehicle, wherein the safety verification is used to determine whether executing the remote driving command will cause the autonomous vending vehicle to collide with an obstacle; and executing the remote driving command when the safety verification passes.

[0016] To address the aforementioned technical problems, the present invention also provides a remote control device for an autonomous vending vehicle, comprising: The data acquisition module is used to acquire the perception data corresponding to the autonomous vending vehicle in real time during the operation of the autonomous vending vehicle. The perception data includes the surrounding environment data of the vehicle, the status data of the goods in the cargo compartment, and the vehicle operation status data. The autonomous vending vehicle and the remote business platform are connected by a dual-link communication. The first link of the dual-link communication is used to transmit autonomous driving instructions and status data, and the second link is used to transmit perception data and control instructions related to remote driving. An event recognition module is used to recognize events based on the perceived data in order to determine whether a preset event is triggered. The preset events include at least stockout events and remote takeover events. The replenishment module is used to generate out-of-stock event information and report it to the remote business platform corresponding to the autonomous vending vehicle when an out-of-stock event is triggered. Then, it receives and executes the replenishment decision instruction issued by the remote business platform in response to the out-of-stock event information, so as to replenish the autonomous vending vehicle. The remote driving module is used to generate and send a remote takeover request to the remote business platform when a remote takeover event is triggered. Subsequently, it receives and executes the remote driving instructions issued by the remote business platform in response to the remote takeover request, so as to remotely control the autonomous vending vehicle.

[0017] To address the aforementioned technical problems, the present invention also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the remote control method for autonomous vending vehicles as described above.

[0018] To address the aforementioned technical problems, the present invention also provides a non-transitory computer-readable storage medium storing computer instructions for causing a computer to execute the remote control method for an autonomous vending vehicle as described above. Attached Figure Description

[0019] Figure 1 This is a flowchart of the remote control method for autonomous vending vehicles according to the present invention; Figure 2 This is a schematic diagram of the remote control device for the autonomous vending vehicle of the present invention; Figure 3 This is a schematic diagram of the electronic device of the present invention. Detailed Implementation

[0020] The principles and features of the present invention are described below. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0021] Example 1 like Figure 1 As shown, this embodiment provides a remote control method for an autonomous vending vehicle, including: S101. During the operation of the autonomous vending vehicle, the perception data corresponding to the autonomous vending vehicle is acquired in real time. The perception data includes data on the surrounding environment of the vehicle, data on the status of goods in the cargo compartment, and data on the vehicle's operating status. The autonomous vending vehicle and the remote business platform are connected by a dual-link communication. The first link in the dual-link communication is used to transmit autonomous driving instructions and status data, and the second link is used to transmit perception data and control instructions related to remote driving.

[0022] S102. Based on the perceived data, perform event identification to determine whether a preset event is triggered. The preset events include at least a stockout event and a remote takeover event.

[0023] S103. When a stockout event is triggered, stockout event information is generated and reported to the remote business platform corresponding to the autonomous vending vehicle. Subsequently, a replenishment decision instruction issued by the remote business platform in response to the stockout event information is received and executed in order to replenish the autonomous vending vehicle.

[0024] S104. When a remote takeover event is triggered, a remote takeover request is generated and sent to the remote business platform. Subsequently, the remote driving command issued by the remote business platform in response to the remote takeover request is received and executed in order to remotely control the autonomous vending vehicle.

[0025] Specifically, data is collected in real time through various sensors, including lidar, cameras, ultrasonic radar, IMU, and high-precision matrix gravity sensors, to obtain real-time road information, information about the vehicle's surrounding environment, and information about the vehicle's own status, thus obtaining perception data, which is then transmitted to the intelligent controller.

[0026] Among them, vehicle surrounding environment data refers to information about the vehicle's external physical space perceived by various sensors. The main sources include: lidar, which generates high-precision 3D point cloud data of the vehicle's surrounding environment by emitting laser beams and receiving reflected signals; cameras, which capture visual images (video streams); and ultrasonic radar, which uses ultrasound for short-range ranging.

[0027] The cargo hold inventory data reflects the stock status of goods carried in the vending vehicle. It mainly comes from high-precision matrix gravity sensors deployed at the bottom of each cargo compartment, collecting time-series data on the pressure (gravity) experienced by each compartment in a grid pattern.

[0028] Vehicle operating status data refers to data reflecting the vehicle's dynamic performance, location, and health status. Main sources include: the inertial measurement unit (IMU) providing information such as vehicle acceleration, angular velocity, and attitude; the global positioning system (GPS) providing global location information such as latitude, longitude, altitude, and speed; and the vehicle controller providing dynamic vehicle information such as vehicle speed, steering angle, remaining battery power, motor status, brake status, and fault codes.

[0029] The intelligent controller communicates with the business platform via a dual-link system, processing information collected from multiple sensors to control the execution of autonomous and remote driving. Based on real-time perception data, the intelligent controller performs event recognition to determine if a stockout or remote takeover event has been triggered. When such an event is triggered, the intelligent controller reports the relevant information to the business platform via the communication link. Once the business platform generates the corresponding instructions, the intelligent controller receives and executes them.

[0030] The intelligent controller and the business platform communicate via a dual-link communication module. The first link is used for autonomous driving, where the intelligent controller uploads real-time data required for autonomous driving, such as vehicle speed, position, and direction, to the business platform, while simultaneously receiving autonomous driving commands from the business platform, such as path planning, speed, and angle. The second link is used for remote driving. When remote intervention is required in special circumstances, the intelligent controller uploads real-time video and sensor data, as well as other information needed for remote driving, to the business platform, and receives remote driving commands from the business platform, such as steering and braking commands.

[0031] Traditional vehicle control systems often employ single-link communication, which is prone to congestion or confusion in information transmission and processing. This method establishes a dual-link communication system: one link is dedicated to transmitting data related to autonomous driving, while the other is used for exchanging information required for remote driving. This clear division of labor ensures that autonomous driving and remote driving functions do not interfere with each other, improving the efficiency and accuracy of information transmission and ensuring that the vehicle can respond quickly to commands under various conditions.

[0032] The dual-link architecture enhances the system's fault tolerance. When one link fails, the other can still maintain some functionality, preventing the vehicle from losing control due to communication interruptions. For example, when signal interference occurs in the autonomous driving link, the remote driving link can still receive vehicle status information, allowing personnel to intervene promptly and ensure vehicle safety.

[0033] The business platform is used to receive information uploaded by the intelligent controller and provide an operation interface for business personnel. Business personnel can perform operations such as replenishment confirmation and remote assistance in vehicle relocation decisions on this platform.

[0034] Under normal circumstances, the autonomous vending vehicle maintains normal operation through the first link. The intelligent controller receives the driving path issued by the business platform and integrates road information collected by sensors and the vehicle's real-time position fed back by the vehicle's inertial navigation system. Based on the preset aiming distance and angle range, it controls the vehicle to drive autonomously along the planned path. At the same time, the vehicle uses sensors such as LiDAR, cameras, and ultrasonic radar to achieve real-time obstacle avoidance and parking operations.

[0035] In addition, the intelligent controller continuously monitors data from the vehicle's gravity sensors and analyzes the inventory status of goods in real time. Once a shortage of a certain type of product is identified, the controller immediately reports the shortage information to the business platform via the autonomous driving link.

[0036] Optionally, in an embodiment, the step of performing event recognition based on the perceived data to determine whether a preset event is triggered includes: extracting pressure time-series data corresponding to each product in the autonomous vending vehicle from the perceived data; for each product, inputting the pressure time-series data corresponding to the product into a trained time-series classification model to identify the product's state category through the time-series classification model; and determining that the out-of-stock event is triggered when any product maintains the out-of-stock state for multiple consecutive determination periods.

[0037] Specifically, a high-precision matrix gravity sensor is deployed in the vehicle's cargo compartment to sense pressure distribution in a grid pattern. The matrix gravity sensor is a single integrated sensor, which can be a monolithic "weighing platform" or "thin-film pressure distribution sensor" specifically tailored to the cargo compartment's dimensions. This sensor itself integrates sensing elements arranged in a matrix or grid pattern. Specifically, the matrix gravity sensor is mounted as a whole on the cargo compartment floor. Its internal array of sensing elements (e.g., a matrix based on strain gauges or capacitive sensing) can simultaneously measure the pressure distribution across the entire plane. The sensor typically integrates a microprocessor that performs preliminary processing on the data from all sensing elements and then directly outputs a complete pressure distribution data matrix via a standard interface (such as I2C or SPI).

[0038] Specifically, in the scenario of autonomous vending vehicles, the noise in the collected stress data mainly comes from: 1. Vehicle vibration and bumps: The continuous high-frequency vibration and low-frequency shaking generated when the vehicle drives over road seams, speed bumps, and uneven road surfaces will directly cause physical deformation and acceleration changes in the cargo compartment and gravity sensors, causing the sensor readings to fluctuate rapidly and frequently around their true values. Even when the vehicle is stationary, the engine's idling vibration will introduce noise.

[0039] 2. Inertial forces generated by vehicle dynamics: When the vehicle accelerates, brakes, or turns, the goods exert additional forces on the sensors due to inertia. During acceleration, the readings may momentarily be higher (similar to being overweight); during braking, the readings may momentarily be lower (similar to weightlessness). When turning, the shift in the center of gravity of the goods also causes uneven forces on the sensors.

[0040] 3. Electrical interference: Electromagnetic interference generated by the motors, controllers, power supplies, and other equipment on the vehicle during operation can couple into the sensor circuitry through the wiring, introducing random, small-amplitude spikes in the readings.

[0041] Without noise reduction, it may lead to missed detections (e.g., the product is actually sold out, but when the vehicle accelerates or goes over a bump, the vibration and inertial force generate a positive pulse, causing the sensor reading to briefly rise above the threshold when unloaded), false detections (e.g., the product is actually still on the shelf, but a violent bump or braking causes the instantaneous reading of the gravity sensor to drop sharply below the out-of-stock threshold), or even the threshold judgment method to fail completely.

[0042] Therefore, to address the noise issue in the data, a sliding window mean filtering algorithm is used to process the pressure data collected by the sensor in order to eliminate noise generated by vehicle vibration.

[0043] Through continuous monitoring by sensors and data preprocessing, pressure time-series data corresponding to each item in the autonomous vending vehicle can be obtained.

[0044] For each product, the corresponding stress time-series data is input into a pre-trained lightweight time-series classification model to identify the product's state category. The model's input is the product's stress time-series data, and the output is the product's state category, specifically including the following four types: STATE_NORMAL (normal), STATE_LOW (low inventory, such as 20% remaining), STATE_EMPTY (out of stock), and STATE_FAULT (sensor malfunction or product tipping over).

[0045] Triggering mechanism: The final stockout event is triggered only when the model outputs STATE_EMPTY for N consecutive cycles (e.g., N=5, each cycle is 3s) to avoid false alarms.

[0046] The lightweight time-series classification model construction process includes: First, data collection and labeling are performed. Raw time-series data streams from the gravity sensors are collected during actual vending vehicle operation. Based on actual replenishment and sales records, status labels are assigned to the data streams. STATE_NORMAL (Normal): Sufficient stock available; STATE_LOW (Low Inventory): The item has approximately 20% remaining. STATE_EMPTY (Out of Stock): The item is sold out. STATE_FAULT (Abnormal): Sensor malfunction or product tipping.

[0047] A fixed-length time window (e.g., 60 time steps, each 1 second) is constructed to generate time-series data on commodity stress, which becomes the model input. The training dataset is then expanded by adding slight Gaussian noise to the original sequence and performing minor time-axis stretching to improve model robustness.

[0048] For the model architecture, a 1D-CNN-based temporal classification model or a lightweight variant based on LSTM is used. In this embodiment, the model architecture is described using a 1D-CNN-based temporal classification model, specifically including: Input layer: (60,1), i.e., 60 time steps, 1 feature dimension; 1D-CNN layer: 16 filters, kernel size = 5, activation function = 'relu'; 1D-CNN layer: 32 filters, kernel size = 3, activation function = 'relu'; Global average pooling layer (replaces fully connected layers, significantly reducing parameters); Dropout layer: rate=0.3 (to prevent overfitting); Fully connected layer: 4 units, activation function = 'softmax' (corresponding to 4 state classifications).

[0049] Based on the aforementioned training data and model architecture, model training and optimization are performed. Furthermore, through knowledge distillation, parameter quantization, and pruning optimization, lightweight model deployment is achieved.

[0050] For time-series classification models, an online optimization mechanism is adopted to achieve continuous learning and online updates, specifically: Feedback loop: When the sales staff confirms the replenishment, the sensor data for that time period is automatically labeled as STATE_EMPTY and added to the training set; Incremental learning: Regularly fine-tune the model using newly collected data to adapt to changes in merchandise display; A / B testing: The new version of the model is first tested on a selection of vehicles to verify its effectiveness before a full update.

[0051] When a stockout event is triggered, the intelligent controller generates and reports the stockout event information to the business platform. The stockout event information uploaded by the intelligent controller is a structured data packet containing: vehicle ID, stockout item ID, stockout timestamp, vehicle's current GPS location, and vehicle's current task status.

[0052] Optionally, in an embodiment, the replenishment decision instruction is obtained by: calculating the comprehensive utility value corresponding to each preset candidate decision scheme based on the stockout event information; selecting the candidate decision scheme with the highest comprehensive utility value from all candidate decision schemes that meet the preset hard constraints as the target decision scheme; and generating the replenishment decision instruction based on the target decision scheme.

[0053] Optionally, in an embodiment, the step of calculating the comprehensive utility value corresponding to each preset candidate decision scheme based on the stockout event information includes: calculating the expected revenue, operating cost, and execution risk corresponding to each preset candidate decision scheme based on the stockout event information; wherein, the expected revenue corresponding to each preset candidate decision scheme represents the expected revenue obtained by executing the preset candidate decision scheme, the operating cost corresponding to each preset candidate decision scheme represents the resource cost required to execute the preset candidate decision scheme, and the execution risk corresponding to each preset candidate decision scheme represents the risk of decreased customer satisfaction or task priority conflict caused by executing the preset candidate decision scheme; identifying the operating scenario of the autonomous vending vehicle based on the stockout event information; adjusting the preset weights according to the operating scenario to obtain new weights; and performing a weighted fusion calculation on each preset candidate decision scheme based on the expected revenue, operating cost, execution risk, and new weights to obtain the comprehensive utility value corresponding to the preset candidate decision scheme.

[0054] The business platform has a built-in replenishment decision engine. Upon receiving a stockout event, this engine does not immediately dispatch vehicles, but instead runs a multi-objective optimization function. Multi-objective optimization can be formalized as a weighted multi-objective decision function, with the goal of finding a decision scheme that maximizes overall utility.

[0055] The multi-objective optimization decision function is: ; in, The overall utility value, The expected return function, For operating cost function, For the penalty function, , and All are weighting coefficients .

[0056] Specifically, the expected return function The expression is: ; in, The expected sales volume of product i in the current region (based on historical data and location); Let i be the unit price of product i; As an urgency factor, out-of-stock items = 1.0, low inventory = 0.3, normal = 0.

[0057] Operating cost function The expression is: =DistanceCost+TimeCost+LaborCost+CoordinationCost; The formula for calculating DistanceCost is as follows: DistanceCost=FuelCost×(ReturnDistance+DetourDistance); FuelCost is the unit price of energy, ReturnDistance is the return distance, and DetourDistance is the detour distance. The formula for calculating TimeCost is: TimeCost=TimeValue×(ReturnTime+DetourTime); TimeValue is the time value coefficient, ReturnTime is the return time, and DetourTime is the detour time; LaborCost is the estimated labor cost, such as the cost of on-site personnel required for processing. CoordinationCost is the scheduling cost (a fixed value) for collaborative replenishment.

[0058] penalty function The expression is: =CustomerDissatisfaction+PriorityViolation; CustomerDissatisfaction is an estimate of dissatisfaction based on the duration of stockouts and regional customer density; PriorityViolation is the penalty for violating high-priority tasks (such as emergency orders).

[0059] In this embodiment, the initial values ​​of each weight are set as follows (which can be adjusted according to business strategy): the expected revenue weight is 0.5, the operating cost weight is 0.3, and the execution risk weight is 0.2.

[0060] Optionally, in an embodiment, the preset weights include expected revenue weight, operating cost weight, and execution risk weight; adjusting the preset weights according to the operating scenario to obtain new weights includes: increasing the value of the expected revenue weight when the operating scenario is that the autonomous vending vehicle is in a target operating period; increasing the value of the execution risk weight when the operating scenario is that the autonomous vending vehicle is in a target operating area; and increasing the value of the operating cost weight when the operating scenario is that the battery power of the autonomous vending vehicle is lower than a preset threshold.

[0061] The context-adaptive dynamic weights are as follows: When the operational scenario involves autonomous vending vehicles operating during the target time period, the weight of expected revenue is increased, while the weights of execution risk and operating cost are correspondingly decreased to ensure that the sum of the three weights equals 1. For example, adjusting the expected revenue weight to 0.7, the operating cost weight to 0.2, and the execution risk weight to 0.1 yields the new weights.

[0062] When the operating scenario involves an autonomous vending vehicle located in the target operating area, the weight of execution risk is increased, while the weights of expected revenue and operating costs are correspondingly decreased to ensure that the sum of the three weights is 1.

[0063] When the battery level of an autonomous vending vehicle falls below a preset threshold during operation, the weight of operating costs is increased, while the weights of expected revenue and execution risk are decreased accordingly, to ensure that the sum of the three weights is 1.

[0064] If none of the above three scenarios are identified, the initial weight values ​​are retained.

[0065] The final output decision variable is {return immediately, return with delay, continue operation}.

[0066] The hard constraints include: 1. Battery Constraint: Before making any decision (especially returning to a refueling point), the vehicle's remaining battery power must be above a minimum safe battery power threshold. If this is not met, options such as "Return Immediately" or "Continue Operation" will be directly excluded, and only the "Emergency Low Battery Handling Procedure" (such as heading to the nearest charging station) can be triggered. The minimum safe battery power threshold can be a dynamically calculated result, for example, it should at least satisfy: the estimated power consumption of heading to the refueling point + safe redundancy battery power.

[0067] 2. Time Window Constraints: Vehicle operation must be limited to a preset working time period. This means satisfying: preset operating hours, fixed working hours of replenishment points, and fixed working hours of back-end monitoring personnel, etc., specifically determined according to requirements.

[0068] 3. Capacity constraint: The total amount of cargo to be replenished must not exceed the maximum physical capacity of the vehicle's cargo hold.

[0069] 4. Priority Task Constraint: This task must be completed first and cannot be interrupted by regular replenishment tasks. For example, in the vehicle's task queue, each task has a "priority" label (e.g., high, medium, low). When making a decision, the optimization engine will first check if there are any currently executing or about-to-be-executed "high" priority tasks. If so, the decision to "immediately return to replenishment" will be temporarily shelved or rejected until the high-priority task is completed.

[0070] The engine outputs the optimal decision: return immediately / delay until the end of this round of tasks / continue operation.

[0071] If the decision is "return immediately," and the business platform issues a restocking task instead of a return to a fixed location, the intelligent controller will invoke the "A+ time window" dynamic path planning algorithm. If traffic congestion or temporary traffic control is encountered during the return journey, the system can replan the optimal route in real time.

[0072] Once the vehicle arrives at the replenishment point, it reports its replenishment status via the autonomous driving link. After the sales personnel complete the physical replenishment and confirm it on the platform, the vehicle receives a new task instruction and resumes normal autonomous driving.

[0073] If the decision is "delay until the end of this round of tasks and return": the current driving path and task plan will not be changed immediately. The vehicle will continue to visit various sales points according to the preset sales route and perform normal autonomous driving and sales tasks. The business platform will not issue a specific return path, but will instead issue a replenishment task with trigger conditions.

[0074] Previously, when autonomous vending vehicles ran out of stock, manual on-site inspections were required to detect it, which was inefficient. In this new method, the intelligent controller monitors the product status in real time and automatically uploads information to the business platform when a stock is out of stock. Business personnel can then directly confirm whether restocking is necessary on the platform and direct the vehicle back to the restocking location. This intelligent processing significantly reduces stockout processing time and improves the operational efficiency of the vending vehicles.

[0075] Optionally, in an embodiment, the step of performing event recognition based on the perceived data to determine whether a preset event is triggered further includes: determining, based on the vehicle operating status data, whether the speed of the autonomous vending vehicle is continuously lower than a speed threshold within a first preset time period; determining, based on the vehicle's surrounding environment data, whether the width of the target passage area of ​​the autonomous vending vehicle is continuously less than a width threshold within a second preset time period; performing a fusion decision to generate a takeover request confidence level when the speed is continuously lower than the speed threshold and the width of the target passage area is continuously less than the width threshold; and triggering the remote takeover event when the takeover request confidence level is greater than a confidence threshold.

[0076] Specifically, when the speed of the autonomous vending vehicle remains below a preset speed threshold for a first preset duration, and the width of the target passage area of ​​the autonomous vending vehicle remains below a preset width threshold for a second preset duration, a fusion decision is executed to generate a takeover request confidence level.

[0077] The intelligent controller integrates LiDAR point cloud data and camera images to construct a real-time map of the passable area around the vehicle using an occupancy grid algorithm. Combining a preset path with inertial navigation data provided by the IMU, a pre-aiming control algorithm calculates the vehicle's required steering angle and speed. The width of the target passable area along the planned path directly in front of the vehicle is extracted from the occupancy grid map and compared to a width threshold.

[0078] Optionally, in an embodiment, the step of performing a fusion decision to generate a takeover request confidence score includes: generating a first sub-confidence score based on the duration of the vehicle speed being lower than a speed threshold and the speed difference between the vehicle speed and the speed threshold; generating a second sub-confidence score based on the duration of the target passage area width being less than a width threshold and the width difference between the target passage area width and the width threshold; and performing a weighted fusion of the first sub-confidence score and the second sub-confidence score based on preset speed condition weights and width condition weights to obtain the takeover request confidence score.

[0079] The duration for which the vehicle speed is below the speed threshold is obtained; that is, the length of time the vehicle's requested speed remains below the speed threshold. The longer this duration, the longer the vehicle remains stationary. The difference between the speed threshold and the actual requested speed is calculated. The larger the difference, the slower the vehicle actually runs, or even the more likely it is to come to a standstill. These two factors are then transformed into a value between 0 and 1 using a mapping function (such as a linear combination, a sigmoid function, or an empirical model based on historical data), which is the first sub-confidence score.

[0080] The duration for which the target passable area width is less than a width threshold is obtained; that is, the length of time the passable area in front of the vehicle remains less than the width threshold. The difference between the width threshold and the target passable area width is calculated. The larger the difference, the narrower the space and the more difficult it is for the vehicle to pass. Similarly, a mapping function method is used to transform these two factors into a value between 0 and 1, namely the second sub-confidence score.

[0081] The takeover request confidence score is calculated by weighting and fusing the first and second sub-confidence scores. A formal remote takeover request is only initiated when the confidence score exceeds a threshold (e.g., 90%), thus reducing the false alarm rate.

[0082] Optionally, in an embodiment, the remote driving command is obtained by: acquiring a scene snapshot data packet for the autonomous vending vehicle; generating a digital twin operating interface for the autonomous vending vehicle and its surrounding environment based on the scene snapshot data packet; calculating and presenting at least one alternative escape route in the digital twin operating interface using a path recommendation algorithm; and generating the remote driving command based on the user's interactive operations on the target location or the alternative escape route in the digital twin operating interface.

[0083] When initiating takeover, a scene snapshot data packet is uploaded via the remote driving link, including: Key video streams: Front-view and rear-view camera video streams encoded with an encoder (e.g., H.265); Structured environmental data: The LiDAR point cloud is downsampled through a voxel grid before being uploaded, which significantly reduces the amount of data. Vehicle status snapshot: The vehicle's current precise pose (from an extended Kalman filter that integrates GPS and IMU), tire swerve, chassis status, etc.

[0084] After receiving the data, the business platform does not simply display the video footage. Instead, it can render and present the vehicle model, the downsampled point cloud environment, and key obstacles (such as pedestrians, vehicles, and other roadblocks) extracted from the video using semantic segmentation algorithms (such as DeepLabV3+) all within a lightweight 3D digital twin interface. This provides remote operators with a depth-based environmental awareness capability that goes beyond 2D video.

[0085] The business platform integrates an escape path recommendation algorithm (such as RRT based on fast randomized tree search). Based on the received environmental data, the algorithm calculates and renders 1 to 3 feasible short-distance escape routes in real time within the digital twin model for the operator's reference.

[0086] Operators no longer directly issue raw steering angles and braking forces. The platform provides a wizard-style operating interface. Operators only need to click on the target point or drag the recommended path in the interface, and the system will automatically convert it into a series of higher-level driving commands (such as MOVE_TO_POINT(x,y), FOLLOW_ARC(radius,angle)) and issue them to the intelligent controller.

[0087] Optionally, in an embodiment, receiving and executing the remote driving command issued by the remote service platform in response to the remote takeover request includes: receiving the remote driving command issued by the remote service platform in response to the remote takeover request; performing a safety verification on the remote driving command based on the local perception data of the autonomous vending vehicle, the safety verification being used to determine whether executing the remote driving command would cause the autonomous vending vehicle to collide with an obstacle; and executing the remote driving command when the safety verification passes.

[0088] After receiving a command, the intelligent controller first verifies it using a local feasibility checker to ensure that the command will not collide with obstacles detected by the real-time sensors. If it is safe, the command is executed; if it is unsafe, a "command rejected" message is sent and a replanning is requested.

[0089] After the vehicle successfully navigated around the obstacle, the remote operator clicked "Resume Autopilot" on the platform. The intelligent controller's state machine automatically switched from "Remote Collaboration Mode" back to "Autopilot Mode" and resumed the original sales task that had been suspended due to the interruption.

[0090] Traditionally, when vehicles encounter insurmountable obstacles, manual intervention is required on-site, consuming significant manpower and time. This new method, through a remote driving link, transmits real-time vehicle video and sensor data to a business platform. Business personnel can remotely determine whether assistance is needed to move the vehicle and issue operational commands. This innovation enables vehicles to quickly overcome obstacles and resume operation, reducing manual intervention and lowering operating costs.

[0091] Furthermore, when handling special situations, business personnel can obtain multi-dimensional information such as real-time vehicle video, sensor data, and connection status through the business platform. This information provides business personnel with comprehensive decision-making basis, enabling them to accurately assess the vehicle's condition and make reasonable decisions. For example, when remotely assisting in moving a vehicle, business personnel can precisely control the vehicle's steering and braking based on camera footage and sensor data to ensure operational safety.

[0092] Through dual-link communication, operational instructions from staff can be transmitted to the intelligent controller in real time, allowing the vehicle to respond immediately. This real-time capability makes remote collaboration more efficient, enabling timely resolution of vehicle issues and ensuring the normal operation of the autonomous vending vehicles.

[0093] This method integrates information from multiple sensor modules, including LiDAR, cameras, and ultrasonic radar. The intelligent controller comprehensively analyzes this information to achieve accurate perception of road and vehicle conditions. Simultaneously, the intelligent controller works closely with the business platform, flexibly adjusting autonomous driving and remote driving strategies based on user instructions and actual vehicle conditions, thus improving the system's intelligence and adaptability.

[0094] In summary, by employing a dual-link communication and remote collaboration mechanism, autonomous vending vehicles can quickly respond and handle special situations such as stockouts or obstacles, effectively reducing vehicle downtime and significantly improving overall operational efficiency. Simultaneously, business personnel can remotely monitor and operate the vehicles through a business platform, drastically reducing the frequency of on-site visits and saving human resources and related costs. Furthermore, this method achieves real-time synchronization of vehicle information and intelligent decision support, comprehensively enhancing the intelligent management level of autonomous vending vehicles and providing reliable technical support for large-scale, efficient commercial operations.

[0095] Example 2 like Figure 2 As shown, this embodiment provides a remote control device 200 for an autonomous vending vehicle, including: The data acquisition module 201 is used to acquire the perception data corresponding to the autonomous vending vehicle in real time during the operation of the autonomous vending vehicle. The perception data includes the surrounding environment data of the vehicle, the status data of the goods in the cargo compartment and the vehicle operation status data. The autonomous vending vehicle and the remote business platform are connected by a dual-link communication. The first link of the dual-link communication is used to transmit autonomous driving instructions and status data, and the second link is used to transmit perception data and control instructions related to remote driving. The event recognition module 202 is used to perform event recognition based on the perceived data to determine whether a preset event is triggered. The preset events include at least a stockout event and a remote takeover event. The replenishment module 203 is used to generate out-of-stock event information and report it to the remote business platform corresponding to the autonomous vending vehicle when an out-of-stock event is triggered. Then, it receives and executes the replenishment decision instruction issued by the remote business platform in response to the out-of-stock event information so as to replenish the autonomous vending vehicle. The remote driving module 204 is used to generate and send a remote takeover request to the remote business platform when a remote takeover event is triggered. Subsequently, it receives and executes the remote driving instructions issued by the remote business platform in response to the remote takeover request, so as to remotely control the autonomous vending vehicle.

[0096] Optionally, in an embodiment, the event recognition module 202 includes: The time-series data acquisition unit is used to extract the pressure time-series data corresponding to each product in the autonomous vending vehicle from the sensing data; The state recognition unit is used to input the pressure time series data corresponding to each product into the trained time series classification model for each product, so as to identify the state category of the product through the time series classification model. The out-of-stock event triggering unit is used to determine and trigger the out-of-stock event when any product continuously maintains the out-of-stock status for multiple consecutive determination periods.

[0097] Optionally, in an embodiment, the event recognition module 202 further includes: The first judgment unit is used to determine, based on the vehicle operating status data, whether the speed of the autonomous vending vehicle is continuously lower than the speed threshold within a first preset time period. The second judgment unit is used to determine, based on the vehicle's surrounding environment data, whether the width of the target passage area of ​​the autonomous vending vehicle is continuously less than a width threshold within a second preset time period. The takeover request confidence calculation unit is used to perform fusion decision to generate a takeover request confidence score when the vehicle speed is continuously lower than the speed threshold and the target passage area width is continuously lower than the width threshold. The remote takeover event triggering unit is used to trigger the remote takeover event when the confidence level of the takeover request is greater than a confidence threshold.

[0098] Optionally, in an embodiment, the takeover request confidence calculation unit includes: The first sub-confidence sub-unit is used to generate a first sub-confidence based on the duration of the vehicle speed being lower than the speed threshold and the speed difference between the vehicle speed and the speed threshold. The second sub-confidence sub-unit is used to generate a second sub-confidence based on the duration during which the target passage area width is less than a width threshold, and the width difference between the target passage area width and the width threshold. The takeover request confidence subunit is used to perform weighted fusion of the first sub-confidence and the second sub-confidence based on preset speed condition weights and width condition weights to obtain the takeover request confidence.

[0099] Optionally, in an embodiment, the replenishment decision instruction is obtained by: calculating the comprehensive utility value corresponding to each preset candidate decision scheme based on the stockout event information; selecting the candidate decision scheme with the highest comprehensive utility value from all candidate decision schemes that meet the preset hard constraints as the target decision scheme; and generating the replenishment decision instruction based on the target decision scheme.

[0100] Optionally, in an embodiment, the step of calculating the comprehensive utility value corresponding to each preset candidate decision scheme based on the stockout event information includes: calculating the expected revenue, operating cost, and execution risk corresponding to each preset candidate decision scheme based on the stockout event information; wherein, the expected revenue corresponding to each preset candidate decision scheme represents the expected revenue obtained by executing the preset candidate decision scheme, the operating cost corresponding to each preset candidate decision scheme represents the resource cost required to execute the preset candidate decision scheme, and the execution risk corresponding to each preset candidate decision scheme represents the risk of decreased customer satisfaction or task priority conflict caused by executing the preset candidate decision scheme; identifying the operating scenario of the autonomous vending vehicle based on the stockout event information; adjusting the preset weights according to the operating scenario to obtain new weights; and performing a weighted fusion calculation on each preset candidate decision scheme based on the expected revenue, operating cost, execution risk, and new weights to obtain the comprehensive utility value corresponding to the preset candidate decision scheme.

[0101] Optionally, in an embodiment, the preset weights include expected revenue weight, operating cost weight, and execution risk weight; adjusting the preset weights according to the operating scenario to obtain new weights includes: increasing the value of the expected revenue weight when the operating scenario is that the autonomous vending vehicle is in a target operating period; increasing the value of the execution risk weight when the operating scenario is that the autonomous vending vehicle is in a target operating area; and increasing the value of the operating cost weight when the operating scenario is that the battery power of the autonomous vending vehicle is lower than a preset threshold.

[0102] Optionally, in an embodiment, the remote driving command is obtained by: acquiring a scene snapshot data packet for the autonomous vending vehicle; generating a digital twin operating interface for the autonomous vending vehicle and its surrounding environment based on the scene snapshot data packet; calculating and presenting at least one alternative escape route in the digital twin operating interface using a path recommendation algorithm; and generating the remote driving command based on the user's interactive operations on the target location or the alternative escape route in the digital twin operating interface.

[0103] Optionally, in an embodiment, the remote driving module 204 includes: The instruction receiving unit is used to receive remote driving instructions issued by the remote service platform in response to the remote takeover request; A safety verification unit is used to perform safety verification on the remote driving command based on the local perception data of the autonomous vending vehicle. The safety verification is used to determine whether executing the remote driving command will cause the autonomous vending vehicle to collide with an obstacle. The instruction execution unit is used to execute the remote driving instruction when the security check passes.

[0104] In some embodiments, the autonomous driving vending vehicle remote control device 200 of the present invention can be implemented in a combination of hardware and software. As an example, the autonomous driving vending vehicle remote control device 200 of the present invention can be a processor in the form of a hardware decoding processor, which is programmed to execute the autonomous driving vending vehicle remote control method of the present invention. For example, the processor in the form of a hardware decoding processor can be one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), or other electronic components.

[0105] The modules described in the embodiments of this invention can be implemented in software or hardware. The names of the modules are not, in some cases, limiting the scope of the module itself.

[0106] Example 3 like Figure 3As shown, this embodiment provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the remote control method for an autonomous vending vehicle as described in Embodiment 1.

[0107] In other words, an electronic device according to an embodiment of the present invention may include, but is not limited to, a processor and a memory; the memory is used to store computer programs; the processor is used to execute the remote control method for autonomous vending vehicles shown in any embodiment of the present invention by calling the computer programs.

[0108] In one alternative embodiment, an electronic device is provided, such as Figure 3 As shown, Figure 3 The illustrated electronic device 300 includes a processor 301 and a memory 303. The processor 301 and the memory 303 are connected, for example, via a bus 302. Optionally, the electronic device 300 may further include a transceiver 304, which can be used for data interaction between the electronic device and other electronic devices, such as sending and / or receiving data. It should be noted that in practical applications, the transceiver 304 is not limited to one type, and the structure of the electronic device 300 does not constitute a limitation on the embodiments of the present invention.

[0109] Processor 301 may be a CPU (Central Processing Unit), a general-purpose processor, a DSP (Digital Signal Processor), an ASIC (Application Specific Integrated Circuit), an FPGA (Field Programmable Gate Array), or other programmable logic devices, transistor logic devices, hardware components, or any combination thereof. It can implement or execute the various exemplary logic blocks, modules, and circuits described in conjunction with the disclosure of this invention. Processor 301 may also be a combination that implements computational functions, such as including one or more microprocessor combinations, a combination of a DSP and a microprocessor, etc.

[0110] Bus 302 may include a path for transmitting information between the aforementioned components. Bus 302 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. Bus 302 can be divided into address bus, data bus, control bus, etc. For ease of representation, Figure 3The bus 302 is represented by only one thick line, but this does not mean that there is only one bus or one type of bus.

[0111] The memory 303 may be a ROM (Read Only Memory) or other type of static storage device capable of storing static information and instructions, RAM (Random Access Memory) or other type of dynamic storage device capable of storing information and instructions, or an EEPROM (Electrically Erasable Programmable Read Only Memory), CD-ROM (Compact Disc Read Only Memory) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0112] The memory 303 is used to store application code (computer program) for executing the present invention, and its execution is controlled by the processor 301. The processor 301 is used to execute the application code stored in the memory 303 to implement the content shown in the foregoing method embodiments.

[0113] Among them, electronic devices can also be terminal devices, which can be any device that can install applications, including at least one of smartphones, tablets, laptops, desktop computers, smart speakers, smartwatches, smart TVs, and smart in-vehicle devices.

[0114] It should be noted that, Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of the present invention.

[0115] Example 4 This embodiment provides a non-transitory computer-readable storage medium that stores computer instructions for causing a computer to execute the remote control method for an autonomous vending vehicle as described in Embodiment 1.

[0116] Alternatively, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a compact disc read-only memory (CD-ROM), magnetic tape, a floppy disk, and an optical data storage device, etc.

[0117] In an exemplary embodiment, a computer program product or computer program is also provided, which includes computer instructions stored in a computer-readable storage medium. A processor of an electronic device reads the computer instructions from the computer-readable storage medium and executes the computer instructions, causing the electronic device to perform the aforementioned remote control method for an autonomous vending vehicle.

[0118] Computer program code for performing the operations of this invention can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as C or similar languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0119] It should be understood that the flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of methods and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0120] The computer-readable storage medium provided in this invention can be, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EEPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0121] The aforementioned computer-readable storage medium carries one or more programs, which, when executed by the electronic device, cause the electronic device to perform the method shown in the above embodiments.

[0122] The above description is merely a preferred embodiment of the present invention and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of disclosure in this invention is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the above-disclosed concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions disclosed in this invention.

[0123] It should be noted that the terms "first," "second," etc., used in the specification and claims of this application are used to distinguish similar objects and represent a limitation on a specific order or sequence. Where appropriate, the order of use for similar objects can be interchanged so that the embodiments of this application described herein can be implemented in an order other than that shown or described.

[0124] Those skilled in the art will recognize that this invention can be implemented as a system, method, or computer program product. Therefore, this invention can be specifically implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this invention can also be implemented as a computer program product contained in one or more computer-readable media, which includes computer-readable program code.

[0125] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.

Claims

1. A remote control method for an autonomous vending vehicle, characterized in that, include: During the operation of the autonomous vending vehicle, the perception data corresponding to the autonomous vending vehicle is acquired in real time. The perception data includes data on the surrounding environment of the vehicle, data on the status of goods in the cargo compartment, and data on the vehicle's operating status. The autonomous vending vehicle and the remote business platform are connected by a dual-link communication. The first link in the dual-link communication is used to transmit autonomous driving commands and status data, and the second link is used to transmit perception data and control commands related to remote driving. Based on the perceived data, event identification is performed to determine whether a preset event is triggered. The preset events include at least stockout events and remote takeover events. When a stockout event is triggered, stockout event information is generated and reported to the remote business platform corresponding to the autonomous vending vehicle. Subsequently, the system receives and executes replenishment decision instructions issued by the remote business platform in response to the stockout event information, so as to replenish the autonomous vending vehicle. When a remote takeover event is triggered, a remote takeover request is generated and sent to the remote business platform. Subsequently, remote driving instructions issued by the remote business platform in response to the remote takeover request are received and executed in order to remotely control the autonomous vending vehicle.

2. The remote control method for autonomous vending vehicles according to claim 1, characterized in that, The step of identifying events based on the perceived data to determine whether a preset event has been triggered includes: Extract the pressure time-series data corresponding to each item in the autonomous vending vehicle from the perceived data; For each product, the pressure time series data corresponding to the product is input into the trained time series classification model so as to identify the state category of the product through the time series classification model; When any product remains in the out-of-stock state for multiple consecutive determination periods, the out-of-stock event is determined to be triggered.

3. The remote control method for autonomous vending vehicles according to claim 1, characterized in that, The step of identifying events based on the perceived data to determine whether a preset event has been triggered further includes: Based on the vehicle operation status data, determine whether the speed of the autonomous vending vehicle remains below a speed threshold within a first preset time period. Based on the vehicle's surrounding environment data, it is determined whether the width of the target passage area of ​​the autonomous vending vehicle is continuously less than a width threshold within a second preset time period. If the vehicle speed remains below the speed threshold and the target passage area width remains below the width threshold, a fusion decision is performed to generate a takeover request confidence level. The remote takeover event is triggered when the confidence level of the takeover request is greater than the confidence threshold.

4. The remote control method for autonomous vending vehicles according to claim 3, characterized in that, The execution of the fusion decision to generate a takeover request confidence level includes: A first sub-confidence level is generated based on the duration of the vehicle speed being below a speed threshold and the speed difference between the vehicle speed and the speed threshold. A second sub-confidence is generated based on the duration during which the target passage area width is less than a width threshold, and the width difference between the target passage area width and the width threshold. Based on preset speed condition weights and width condition weights, the first sub-confidence and the second sub-confidence are weighted and fused to obtain the takeover request confidence.

5. The remote control method for an autonomous vending vehicle according to claim 1, characterized in that, The replenishment decision instruction is obtained in the following way: Based on the stockout event information, calculate the comprehensive utility value corresponding to each preset candidate decision scheme; From all candidate decision schemes that meet the preset hard constraints, select the candidate decision scheme with the highest comprehensive utility value as the target decision scheme; Based on the target decision scheme, the replenishment decision instruction is generated.

6. The remote control method for an autonomous vending vehicle according to claim 5, characterized in that, The step of calculating the comprehensive utility value corresponding to each preset candidate decision scheme based on the stockout event information includes: Based on the stockout event information, calculate the expected revenue, operating costs, and execution risks for each preset candidate decision-making scheme; Among them, the expected revenue corresponding to each preset candidate decision scheme represents the expected revenue obtained by executing the preset candidate decision scheme, the operating cost corresponding to each preset candidate decision scheme represents the resource cost required to execute the preset candidate decision scheme, and the execution risk corresponding to each preset candidate decision scheme represents the risk of decreased customer satisfaction or task priority conflict caused by executing the preset candidate decision scheme. Based on the out-of-stock event information, identify the operating scenario of the autonomous vending vehicle; Based on the operational scenario, the preset weights are adjusted to obtain new weights; For each preset candidate decision scheme, a weighted fusion calculation is performed based on the expected benefits, operating costs, execution risks, and new weights corresponding to the preset candidate decision scheme to obtain the comprehensive utility value corresponding to the preset candidate decision scheme.

7. The remote control method for an autonomous vending vehicle according to claim 6, characterized in that, The preset weights include expected return weights, operating cost weights, and execution risk weights; adjusting the preset weights according to the operating scenario to obtain new weights includes: When the operating scenario is that the autonomous vending vehicle is in the target operating period, the value of the expected revenue weight is increased; When the operational scenario is that the autonomous vending vehicle is located in the target operational area, the value of the execution risk weight is increased; When the operating scenario is that the battery charge of the autonomous vending vehicle is lower than a preset threshold, the value of the operating cost weight is increased.

8. The remote control method for an autonomous vending vehicle according to claim 1, characterized in that, The remote driving command is obtained in the following way: Obtain scene snapshot data packets for the autonomous vending vehicle; Based on the scene snapshot data package, a digital twin operating interface is generated for the autonomous vending vehicle and its surrounding environment. In the digital twin operation interface, at least one alternative escape path is calculated and presented through a path recommendation algorithm; The remote driving command is generated based on the user's interactive operations on the digital twin operating interface regarding the target location or the alternative escape route.

9. The remote control method for an autonomous vending vehicle according to claim 1, characterized in that, The receiving and execution of remote driving instructions issued by the remote service platform in response to the remote takeover request includes: Receive remote driving instructions issued by the remote service platform in response to the remote takeover request; Based on the local perception data of the autonomous vending vehicle, a safety verification is performed on the remote driving command. The safety verification is used to determine whether executing the remote driving command will cause the autonomous vending vehicle to collide with an obstacle. When the security check passes, the remote driving command is executed.

10. A remote control device for an autonomous vending vehicle, characterized in that, include: The data acquisition module is used to acquire the perception data corresponding to the autonomous vending vehicle in real time during the operation of the autonomous vending vehicle. The perception data includes the surrounding environment data of the vehicle, the status data of the goods in the cargo compartment, and the vehicle operation status data. The autonomous vending vehicle and the remote business platform are connected by a dual-link communication. The first link of the dual-link communication is used to transmit autonomous driving instructions and status data, and the second link is used to transmit perception data and control instructions related to remote driving. An event recognition module is used to recognize events based on the perceived data in order to determine whether a preset event is triggered. The preset events include at least stockout events and remote takeover events. The replenishment module is used to generate out-of-stock event information and report it to the remote business platform corresponding to the autonomous vending vehicle when an out-of-stock event is triggered. Then, it receives and executes the replenishment decision instruction issued by the remote business platform in response to the out-of-stock event information, so as to replenish the autonomous vending vehicle. The remote driving module is used to generate and send a remote takeover request to the remote business platform when a remote takeover event is triggered. Subsequently, it receives and executes the remote driving instructions issued by the remote business platform in response to the remote takeover request, so as to remotely control the autonomous vending vehicle.

Images