Emergency response dispatching method and system based on unmanned aerial vehicle, bus and taxi

Abstract

The application discloses an emergency response scheduling method and system based on a UAV, a bus and a taxi, and belongs to the field of emergency scheduling. The application analyzes an emergency response mode of the UAV, establishes an energy consumption model of the UAV under the joint space-time constraints of the bus and the taxi, proposes a participating relay taxi recruitment model, establishes a joint cost optimization model considering response delay, response duration and relay vehicle cost, designs a neural network module based on supervised learning to mine a daily travel model of the taxi, and establishes a bus recruitment model considering the taxi relay. A non-overlapping joint coverage gain greedy algorithm is designed to select appropriate buses to carry the UAV from all bus line sets, effectively balancing the urban emergency space-time coverage and space-time grid coverage benefits. The application provides an efficient and low-cost solution for urban emergency response, and has significant application value.

Description

Technical Field

[0001] This invention relates to the field of low-altitude economic technology, and in particular to an emergency response dispatching method and system based on drones, buses, and taxis. Background Technology

[0002] Advanced Air Mobility (AAM) aims to integrate innovative aviation technologies (small drones and large air taxis) with intelligent transportation management systems to provide cities and regions with more efficient, safe, and environmentally friendly on-demand aviation use cases. A key application scenario for AAM is urban emergency response, offering new possibilities for low-cost and efficient urban emergency response mechanisms. Drones are an important component of AAM, attracting widespread attention due to their flexibility and low cost. Numerous scholars have explored the application of drones in various urban scenarios; however, limited by battery capacity, drones can only perform emergency missions within their flight time.

[0003] Current research addresses the drone's endurance issue primarily through two approaches: internal and external energy sources. Internal energy sources mainly refer to improving battery capacity and efficiency. External energy sources include equipping drones with solar panels to utilize sunlight for additional energy, deploying fixed charging stations, and using mobile ground charging vehicles (MCVs) to replenish drone power, thereby extending their flight time. However, solar panels limit drones' emergency response capabilities at night, city-wide charging station deployments are prohibitively expensive, and the dispatching of numerous MCVs introduces additional traffic congestion. Existing solutions not only waste significant amounts of power chasing ground charging vehicles but also pose flight safety risks during long-distance flights. Therefore, integrating existing urban public transportation systems to provide drones with a safer, more efficient, and flexible emergency response model is crucial. Summary of the Invention

[0004] This invention provides an emergency response dispatching method and system based on drones, buses, and taxis. By mounting drones on the roof of buses for charging and inspection, and coordinating with surrounding taxis to respond to unpredictable random emergencies in cities, the flexibility and efficiency of urban emergency response are improved without introducing new means of transportation or occupying public infrastructure resources.

[0005] A first aspect of the present invention provides an emergency response dispatch method based on drones, buses, and taxis, comprising the following steps:

[0006] Step 1: Model the mobility of buses and taxis based on actual vehicle trajectory data to obtain vehicle mobility models;

[0007] Step 2: Calculate the energy consumption of the UAV during the emergency response based on the vehicle movement model, and analyze the UAV emergency response mode under various relay conditions from the perspective of energy consumption, and construct an energy consumption model for the emergency response process of a single UAV.

[0008] Step 3: Establish a taxi recruitment model to quantify the cost of relay taxis;

[0009] Step 4: Based on the energy consumption model of the emergency response process of the individual UAV and the relay taxi cost calculated by the taxi recruitment model, and taking into account the UAV response delay, UAV response duration and relay taxi recruitment cost, establish a joint cost model for UAV emergency response.

[0010] Step 5: Optimize the aforementioned joint cost model for UAV emergency response and select a target UAV to complete the emergency mission;

[0011] Step 6: Extend the emergency response dispatch of a single drone to emergency response dispatch of multiple drones. Use the taxi travel demand prediction model to predict taxi occupancy and travel time, and quantify the coverage performance of taxis in the city.

[0012] Step 7: Based on the joint cost model of drone emergency response for individual buses and taxis in Step 5 and the taxi travel demand prediction model in Step 6, calculate the spatiotemporal coverage of a single bus route.

[0013] Step 8: Based on the spatiotemporal coverage of the single bus route, use the non-overlapping joint coverage gain greedy algorithm to select the optimal set of buses equipped with drones from all bus route sets, and determine the set of bus routes equipped with drones.

[0014] Optionally, in one embodiment of the present invention, step 1 specifically includes:

[0015] Discretize the entire city into a grid with the same side length. Given that the bus route and trajectory are known, the trajectory of any bus b is defined as follows: in, A tuple representing the grid ID and timestamp, i.e. (g k ,t k ), where k is the total number of trajectory points;

[0016] The time when taxis participate in drone relay operations is divided into a series of time slots. Each time slot has the same duration. Only unoccupied taxis are considered for completing the drone relay mission, let N... un (g i ,s j ) indicates a time slot Period grid The number of unoccupied taxis for any timestamp Occurring in the grid In the event of an emergency, a binary variable is defined to indicate whether a relay taxi is carrying a drone from bus b.

[0017]

[0018] Wherein, r(g n ,t,v)=1 indicates that drones are allowed to operate from the grid containing the taxi grid. Unoccupied taxis were used to transport passengers to emergency grid areas.

[0019] Optionally, in one embodiment of the present invention, the takeoff of the drone from the bus to the emergency point is defined as the outbound journey, and the return of the drone to the original bus after completing the mission is defined as the return journey. In step 2, the analysis of drone emergency response modes under various relay conditions from the perspective of energy consumption includes:

[0020] There are four scenarios based on whether there are relay vehicles on the outbound and return journeys: Scenario 1: No relay vehicles on either the outbound or return journey; Scenario 2: Relay vehicles on the outbound journey but no relay vehicles on the return journey; Scenario 3: No relay vehicles on the outbound journey but relay vehicles on the return journey; Scenario 4: Relay vehicles on both the outbound and return journeys.

[0021] drone trajectory Represented as:

[0022]

[0023] Among them, when drone u responds to emergency G at timestamp t0 p At that time, the flight trajectory was redefined G1(b) represents the grid where the bus is located at this timestamp, and the grid where the emergency point is located is G. p , express The encounter grid between the drone and the bus under the given conditions, where v is the relay taxi.

[0024] Optionally, in one embodiment of the present invention, step 2, constructing an energy consumption model for the emergency response process of a single UAV, includes:

[0025] In scenario 1, the total energy consumption of the drone for:

[0026]

[0027] Wherein, α1 and α2 represent the power consumption of the UAV during flight and when hovering over the emergency point, respectively. For the hovering time of the drone at the emergency point, Dis(G1(b),G p () is a straight path. This represents the straight-line path from the emergency point to the bus grid after the drone completes its mission;

[0028] Select the grid that provides the maximum response time for the drone as the final encounter grid between the drone and the bus.

[0029]

[0030] in, Let t be the average speed of the drone. meet The final encounter between the drone and the bus is represented by t0, which is the moment the emergency mission occurs. The time it takes for the drone to reach the emergency grid. E represents the remaining battery power of the drone before it performs an emergency mission, where λ is the percentage of the drone's battery capacity. c For drone battery capacity;

[0031] In scenario 2, the total energy consumption of the drone for:

[0032]

[0033] Where ε is a default constant;

[0034] The final encounter between the drone and the bus (grid) for:

[0035]

[0036] In scenario 3, the total energy consumption of the drone for:

[0037]

[0038] Unlike scenarios 1 and 2, the drone's hovering time In addition to energy consumption constraints, we also need to consider whether there are available taxis in the emergency point grid, and the final encounter grid between the drone and the bus. for:

[0039]

[0040] in, The travel time for a taxi from the emergency point grid to the meeting grid; Bus B arrives at the grid Timestamps, based on the expected bus trajectory Sure; Indicates drone hovering Afterwards, unoccupied taxis in the emergency response point grid can be used as relay vehicles;

[0041] In scenario 4, the total energy consumption of the drone for:

[0042]

[0043] The taxi and the bus eventually met at the grid. for:

[0044]

[0045] in, This refers to the travel time for a taxi to travel from the emergency point grid to the meeting grid. Bus B arrives at the grid Timestamp.

[0046] Optionally, in one embodiment of the present invention, step 3 specifically includes:

[0047] Based on the routes and passenger occupancy information uploaded by taxis, calculate separately The cost of recruiting relay taxis for both outbound and return trips is rewarded based on the Euclidean distance between the taxi's departure and arrival grids. The taxi relay cost is calculated as follows:

[0048]

[0049] Among them, R taxi Taxi reward per unit length.

[0050] Optionally, in one embodiment of the present invention, step 4 specifically includes:

[0051] According to any bus UAV i Response delays under different drone emergency response modes Response duration and the cost of recruiting relay taxis Establish an emergency utility function for a single bus:

[0052]

[0053] in, for The joint emergency cost of getting off bus B; and The maximum values ​​for response delay, response duration, and relay taxi recruitment cost, respectively; ω i It is the weight corresponding to the cost;

[0054] Establish a joint cost model for emergency response using drones, and determine the cost of bus b. i Ultimate emergency effectiveness The maximum emergency effectiveness under the four drone emergency response modes:

[0055]

[0056] Optionally, in one embodiment of the present invention, step 5 specifically includes:

[0057] Taking into account the emergency response delays, duration, and costs associated with drones, the emergency task was assigned to bus b, which ultimately has the greatest emergency effectiveness. i Drones on the surface:

[0058]

[0059] in, Bus b at timestamp t0 n The number of drones.

[0060] Optionally, in one embodiment of the present invention, step 6 specifically includes:

[0061] A taxi travel demand prediction model is built using a neural network model to predict taxi passenger occupancy and the time required for a taxi to travel from one grid to another.

[0062] Optionally, in one embodiment of the present invention, step 7 specifically includes:

[0063] Based on the emergency utility function of a single bus, if the drone on bus b can respond to g... r In an emergency, otherwise, for The coverage model for a single bus is represented as follows:

[0064]

[0065] in, The larger the bus b i The greater the spatiotemporal coverage effect of drones;

[0066] Considering cost constraints, select some buses to be equipped with drones. The goal is to maximize the selected bus set while keeping the total cost R within the budget B. The combined coverage effect:

[0067]

[0068] stR≤B

[0069] in, Collect the selected buses The combined coverage effect.

[0070] A second aspect of the present invention provides an emergency response dispatch system based on drones, buses, and taxis, comprising:

[0071] The mobility modeling module is used to model the mobility of buses and taxis based on actual vehicle trajectory data to obtain vehicle mobility models.

[0072] The energy consumption model construction module is used to calculate the energy consumption of the UAV during the emergency process based on the vehicle movement model, and to analyze the UAV emergency response mode under various relay conditions from the perspective of energy consumption, and to construct an energy consumption model for the emergency response process of a single UAV.

[0073] The relay cost calculation module is used to establish a taxi recruitment model and quantify the cost of relay taxis.

[0074] The joint cost establishment module is used to establish a joint cost model for drone emergency response based on the energy consumption model of the emergency response process of the individual drone and the relay taxi cost calculated by the taxi recruitment model, taking into account the drone response delay, drone response duration and relay taxi recruitment cost.

[0075] The optimization module is used to optimize the emergency joint cost model of the UAV and select the target UAV to complete the emergency mission;

[0076] The prediction module is used to extend the emergency response scheduling of a single drone to the emergency response scheduling of multiple drones. It uses a taxi travel demand prediction model to predict taxi occupancy and travel time, and quantifies the coverage performance of taxis in the city.

[0077] The coverage module is used to calculate the spatiotemporal coverage of a single bus route based on the joint cost model of drone emergency response for individual buses and taxis, as well as the taxi travel demand prediction model.

[0078] The scheduling module is used to select the optimal set of buses equipped with drones from all bus route sets based on the spatiotemporal coverage of the single bus route using a greedy algorithm with non-overlapping joint coverage gain, thereby determining the set of bus routes equipped with drones.

[0079] This invention presents an emergency response dispatching method and system based on drones, buses, and taxis. The method analyzes the emergency response mode of drones and establishes an energy consumption model for the drone emergency response process under the spatiotemporal constraints of combined bus and taxi trajectories. Then, a participatory relay taxi recruitment model is proposed. Based on this, a joint cost model for drone emergency response considering response delay, response duration, and relay vehicle costs is established. Furthermore, to adapt to the time-varying characteristics of taxi distribution, a supervised learning-based neural network module is designed to mine daily taxi travel patterns, and a bus recruitment model considering taxi relay is established. A greedy algorithm with non-overlapping joint coverage gain is used to select the optimal set of buses carrying drones from all bus route sets, thus determining the bus routes carrying drones. The beneficial effects of this invention are as follows:

[0080] (1) A novel multimodal transport emergency dispatch framework based on buses, taxis, and drones is proposed to dynamically respond to large-scale urban emergency events. This framework can effectively respond to urban emergency needs during the AAM transition phase. The innovative collaborative air-ground coordination of drones, buses, and taxis in urban emergency response overcomes the problem of short response time when using only buses equipped with drones.

[0081] (2) Considering the dynamic distribution of taxis, a data-driven taxi state estimation algorithm was designed. Taking into account the response delay of UAVs, response duration and the cost of recruiting relay taxis, a bus recruitment algorithm based on the joint coverage performance of UAV buses and taxis was proposed to cope with unpredictable emergency response events in the city.

[0082] (3) The performance of the proposed emergency dispatch strategy is comprehensively evaluated using a large-scale urban vehicle trajectory dataset.

[0083] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0084] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein:

[0085] Figure 1 This is a schematic diagram of the architecture of an emergency response dispatching method based on drones, buses, and taxis according to an embodiment of the present invention;

[0086] Figure 2 This is a schematic diagram illustrating taxi travel demand prediction and bus recruitment according to an embodiment of the present invention;

[0087] Figure 3This is a schematic diagram illustrating the impact of the proportion of buses on the emergency coverage area according to an embodiment of the present invention.

[0088] Figure 4 This is a schematic diagram illustrating the impact of the proportion of buses on the spatiotemporal coverage rate in an embodiment of the present invention.

[0089] Figure 5 This is a schematic diagram illustrating the impact of the number of buses on the emergency coverage area according to an embodiment of the present invention;

[0090] Figure 6 This is a schematic diagram illustrating the impact of the number of buses on the spatiotemporal coverage rate according to an embodiment of the present invention.

[0091] Figure 7 This is a schematic diagram illustrating the impact of time period on the emergency coverage area according to an embodiment of the present invention;

[0092] Figure 8 This is a schematic diagram illustrating the impact of time period on the spatiotemporal coverage of an embodiment of the present invention;

[0093] Figure 9 This is a schematic diagram illustrating emergency delays under different ERD requirements according to an embodiment of the present invention;

[0094] Figure 10 This is a schematic diagram illustrating the hovering time under different ERD requirements in an embodiment of the present invention;

[0095] Figure 11 This is a schematic diagram of spatiotemporal coverage under different ERD requirements in an embodiment of the present invention;

[0096] Figure 12 This is a schematic diagram illustrating the required number of buses under different spatiotemporal coverage requirements according to an embodiment of the present invention;

[0097] Figure 13 This is a schematic diagram illustrating the vehicle recruitment costs required under different spatiotemporal coverage requirements according to an embodiment of the present invention.

[0098] Figure 14 This diagram illustrates the required infrastructure costs under different spatiotemporal coverage requirements according to an embodiment of the present invention. Detailed Implementation

[0099] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings, wherein the same or similar reference numerals denote the same or similar elements or elements having the same or similar functions throughout. The embodiments described below with reference to the accompanying drawings are exemplary and intended to explain the present invention, and should not be construed as limiting the present invention.

[0100] like Figure 1 As shown, the emergency response process for a drone riding a bus includes the following four steps.

[0101] 1. Normally, buses are equipped with various types of emergency drones on their roofs, such as emergency medical delivery, reconnaissance and surveillance, communication relay, and search and rescue drones, which are on standby at any time.

[0102] 2. Upon receiving a mission, the corresponding type of drone can fly to the emergency site in two ways: 1. Take off immediately from the bus and fly to the emergency site. 2. Take off from the bus, land on a nearby taxi (if the emergency site is too far from the bus or the drone's current battery level is insufficient for ERD requirements), and then the drone will travel by taxi to the emergency site.

[0103] 3. Upon arrival, the drone hovers over the emergency area, performing response tasks. For example, a drone can deliver first-aid kits to the scene of a traffic accident, or its camera can capture and transmit video footage to assist rescue teams.

[0104] 4. After the emergency is resolved or the drone battery is about to run out, the drone can choose to fly directly back to the original bus, or take a taxi near the emergency site back to the original bus to recharge and continue to complete the next emergency mission.

[0105] To realize emergency response drones based on bus-coordinated relay taxis, the following prerequisites are proposed:

[0106] 1. The takeoff and landing schedule of the drone should not affect the normal mobility of buses and taxis. Furthermore, the drone should be equipped with a visual landing system to achieve precise takeoff and landing on the roofs of buses and taxis.

[0107] 2. Buses and taxis need to be equipped with mobile data connectivity (cellular networks) and other communication devices for communication between drones, buses, taxis, and emergency response control centers. Buses and taxis need to report their trajectory information in real time so that drones can adjust their flight paths to coordinate with them.

[0108] 3. To ensure the efficiency of emergency response, all taxis that meet the emergency response conditions must fully comply with emergency response dispatch, with a compliance rate of 100%.

[0109] like Figure 1 As shown, this emergency response dispatch method based on drones, buses, and taxis includes the following steps:

[0110] Step 1: Model the mobility of buses and taxis based on actual vehicle trajectory data to obtain vehicle mobility models.

[0111] Step 2: Calculate the energy consumption of the UAV during the emergency response based on the vehicle movement model, and analyze the UAV emergency response mode under various relay conditions from the perspective of energy consumption, and construct an energy consumption model for the emergency response process of a single UAV.

[0112] Step 3: Establish a taxi recruitment model and quantify the cost of relay taxis.

[0113] Step 4: Based on the energy consumption model of the emergency response process of a single UAV and the relay taxi cost calculated by the taxi recruitment model, and taking into account the UAV response delay, UAV response duration and relay taxi recruitment cost, establish a joint cost model for UAV emergency response.

[0114] Step 5: Optimize the joint cost model for emergency drone operations and select the target drone to complete the emergency mission.

[0115] Step 6: Extend the emergency response dispatching of single drones to emergency response dispatching of multiple drones. Utilize the taxi travel demand prediction model to predict taxi occupancy and travel time, and quantify the coverage performance of taxis in the city.

[0116] Step 7: Based on the joint cost model of drone emergency response for individual buses and taxis in Step 5 and the taxi travel demand prediction model in Step 6, calculate the spatiotemporal coverage of a single bus route.

[0117] Step 8: Based on the spatiotemporal coverage of a single bus route, use a greedy algorithm with non-overlapping joint coverage gain to select the optimal set of buses equipped with drones from all bus route sets, thus determining the set of bus routes equipped with drones.

[0118] In the embodiments of this invention, steps 1 to 4 are modeling processes, and step 5 optimizes the model from step 4 to select the most suitable drone on a bus for rescue. Steps 1 to 5 represent modeling and optimization of the single drone's response process to an emergency. Step 6 requires expansion to multiple drone rescue. This expansion quantifies the spatiotemporal coverage of multiple buses carrying drones in the city. However, this coverage needs to consider taxi relay capacity; therefore, a taxi demand prediction model is used to predict taxi activity around the buses carrying drones and near emergency locations. In step 7, combining the predicted taxi occupancy and travel time of the OD grid from step 6 with the single drone joint cost optimization model from step 5, the spatiotemporal coverage of a single bus route considering relay taxis can be calculated. In step 8, based on the spatiotemporal coverage of a single bus route calculated in step 7, and considering the bus quantity budget, a non-overlapping joint coverage gain greedy algorithm is proposed to select the most suitable bus route from all bus route sets that satisfies the bus quantity budget, aiming to determine which buses can be used to carry drones.

[0119] In one embodiment of the invention, in step 1, the entire city is discretized into a grid with the same side length (e.g., 1 kilometer). Assuming the bus route and trajectory are known, for any bus b, its trajectory is defined as follows: in, A tuple representing the grid ID and timestamp, i.e. (g k ,t k ), where k is the total number of trajectory points. Due to the diversity of bus speeds, the future trajectories of buses need to be estimated periodically (e.g., every half hour). The trajectory estimation work is the responsibility of the city department. The time during which taxis participate in drone relay operations (from 0:00 to 24:00) is divided into a series of time slots. Each time slot has the same duration, for example, 1 minute. To ensure the effectiveness of emergency response missions and to avoid impacting the taxi passenger experience, only unoccupied taxis (those without passengers) are considered for drone relay missions. Let N un (g i ,s j ) indicates a time slot Period grid The number of unoccupied taxis. For any timestamp Occurring in the grid An emergency occurred at the location. A binary variable was defined to indicate whether a relay taxi carrying a drone on bus b existed.

[0120]

[0121] Wherein, r(g n ,t,v)=1 indicates that the drone can be generated from the grid containing the taxi grid. Unoccupied taxis were used to transport passengers to emergency grid areas.

[0122] In one embodiment of the present invention, in step 2, an energy consumption model for the UAV response process is established. This model is divided into four scenarios based on whether a relay vehicle is involved in the two processes: the UAV taking off from the bus and arriving at the emergency point (hereinafter referred to as the outbound journey), and the UAV returning to the original bus after completing its mission (hereinafter referred to as the return journey). The model then presents seven flight processes under these four scenarios for the UAV to complete its response mission:

[0123] Scenario 1, no relay vehicles involved on either the outbound or return journey: After the UAV control station receives or predicts an emergency, it assigns the response task to the most suitable available UAV, taking into account response latency, response duration, and relay vehicle costs. The available UAV will take off from the bus and fly directly to the mission point, as in process (1). Assuming the future trajectory of the bus is known, let G1(b) represent the grid where the bus is located at the indentation, and let the grid where the emergency point is located be G. p Let Dis(G) O G D )express and The Euclidean distance between the geometric centers of the two grids. Upon arrival at the emergency site, the drone hovers over the site (e.g., for traffic accident monitoring or medical supply delivery), as part of process (2). and They represent in The time it takes for the drone to reach the emergency grid and the hovering time at the emergency point (hereinafter referred to as response duration). Hovering Afterwards, the drone returned to the bus, as part of process (3). express A grid showing the encounter between the drone and the bus under certain conditions. Note the drone's return flight distance onto the bus. and related.

[0124] Scenario 2, with a relay vehicle involved on the outbound journey: The drone takes off from the bus and encounters a taxi within grid G1(b), as part of process (4). After the encounter, the taxi carries the drone to the emergency point grid G. p As shown in process (5). Drone hovering Then return to the bus, repeating the same process (3). The drone's flight distance... Timestamps of taxi arrival times at emergency point grid and the hovering time of the drone related. Determined using a data-driven approach. Determined based on the future trajectory of buses and the energy consumption model of drones.

[0125] Scenario 3, with relay vehicles involved on the return trip: The process of the drone taking off from the bus and arriving at the emergency point grid is the same as (1). Hovering Afterwards, the drone carries a taxi within the emergency point grid back to the original bus as process (6) and process (7). The distance the taxi travels to the grid where it meets the bus is... Its and And it relates to the future trajectory of the bus. In other words, because the drones hover at emergency points for different durations, the grid G ​​where the taxi and bus meet... e different.

[0126] Scenario 4, with relay vehicles involved on both the outbound and return journeys: This process combines scenarios 2 and 3, and will not be discussed in detail here to avoid unnecessary repetition.

[0127] When the drone u responds to the emergency G at timestamp t0 p At that time, the flight trajectory was redefined Where G0 is the grid where the drone leaves the bus (G1(b) in case 1). Combining the flight processes of the drone under cases 1 to 4, the drone trajectory... It can be represented as:

[0128]

[0129] make This represents the energy consumption required for the drone to complete the m-th emergency event under scenario i, with its battery fully charged. To ensure the drone can handle unforeseen circumstances such as communication failures or public transportation delays, the drone's energy consumption should meet the following requirements:

[0130]

[0131] Among them, E c This indicates the total battery capacity of the drone; This represents the remaining energy before responding to the m-th task; λ (e.g., 10%) represents the percentage of energy consumed by the drone in response to emergencies.

[0132] The details of the power consumption of drones in four emergency modes are as follows:

[0133] Case 1, It is a straight path. This indicates that the drone is hovering at the emergency point. This represents the straight-line path from the emergency point to the bus grid after the mission is completed. The total energy consumption of the drone can be calculated. As shown in equation (3):

[0134]

[0135] Here, α1 and α2 represent the power consumption of the UAV during flight and when hovering over the emergency point, respectively; α1 depends not only on the weight of the UAV but also on external factors such as wind speed and air density; while α2 mainly depends on the weight of the UAV. Based on the expected trajectory points of the bus, the UAV needs to continuously estimate the grid at the emergency point where it will meet the bus. Since the required hovering time of the UAV in an emergency scenario cannot be determined in advance, the hovering time of the UAV should be as long as possible. The grid that can provide the maximum response duration for the UAV is selected as the final meeting grid between the UAV and the bus.

[0136]

[0137] in, This is the average speed of the drone; set it to the default value.

[0138] In scenario 2, This represents the trajectory of the drone encountering a taxi within G1(b). Since the unoccupied taxi could get close enough to the bus, therefore... The flight distance can be assumed to be a default constant ε (e.g., 0.1 km). This indicates that the drone travels to the emergency location in a taxi, during which the drone consumes no energy. The rest of the process is similar to Case 1; the total energy consumption of the drone can be calculated. As shown in the equation, and the final encounter grid between the drone and the bus. As shown in equation (7):

[0139]

[0140] In scenario 3, This indicates that after the drone completes its emergency mission, it will connect with the emergency point grid G. p The trajectory of taxis meeting within the vehicle. This indicates that the drone carrying a taxi and a bus encountered each other, and the process consumed no energy. This indicates that after the taxi reaches the encounter grid, the drone flies back to the original bus trajectory from above the taxi. Since the taxi is not in use, it can get close enough to the emergency point and the bus. Therefore, the path... and path All are set to the default constant ε. The total energy consumption of the drone can then be calculated. As shown in equation (11):

[0141]

[0142] Unlike scenarios 1 and 2, the drone's hovering time In addition to energy consumption constraints, it is also necessary to consider whether there are available taxis in the emergency point grid. Therefore, the final encounter grid between the drone and the bus is calculated as shown in equation (12).

[0143]

[0144] in, Indicates drone hovering Subsequently, unoccupied taxis in the emergency point grid can serve as relay vehicles. Constraints require that taxis must arrive at the meeting grid before the shared vehicle. in, This represents the travel time of a taxi from the emergency point grid to the meeting grid, determined based on a data-driven algorithm; This represents bus b arriving at the grid. Timestamps, based on the expected bus trajectory Sure.

[0145] Case 4 combines elements of Cases 2 and 3; to avoid unnecessary repetition, the routes of the drone and taxi will not be described again. The total energy consumption of the drone can be calculated. And the maximum hovering time of drones and the grid of final encounter between taxis and buses

[0146]

[0147] In one embodiment of the present invention, in step 3, a taxi recruitment model is established. Inspired by audience-based taxi research, the model is calculated based on the routes and passenger occupancy information uploaded by taxis. The cost of recruiting relay taxis for both the outbound and return journeys. A reward is given based on the Euclidean distance between the taxi's departure and arrival grids. Therefore, the relay cost of taxis can be calculated:

[0148]

[0149] Among them, R taxi Taxi incentives per unit length, for example, 3 RMB / km.

[0150] In one embodiment of the present invention, in step 4, according to equations (1)-(21), any bus can be obtained. UAV i Response delays under different response conditions (Case 1-Case 4) Response duration and relay costs Taking into account the above factors, an emergency utility function for a single bus is established:

[0151]

[0152] in, for The joint emergency cost of getting off bus B; and ω represents the maximum values ​​for response delay, response duration, and relay cost (default is a fixed value). i This refers to the weighting of the corresponding costs. Bus b i Ultimate emergency effectiveness The maximum emergency utility under four response scenarios:

[0153]

[0154] In one embodiment of the present invention, in step 5, for the timestamp exist An emergency occurring at a certain point can be represented by a set of buses that can cover that point. Taking into account the delays, duration, and cost of emergency response from drones, this task is assigned to bus b, which will ultimately have the greatest emergency effectiveness. i Drones on the surface:

[0155]

[0156] in, Bus b at timestamp t0 n Set the number of drones to a fixed value, such as 5.

[0157] In one embodiment of the present invention, in step 6, a taxi travel demand prediction model is built using a neural network model to predict the passenger occupancy of taxis and the time required for a taxi to travel from one grid to another.

[0158] Specifically, a key challenge in multi-drone scheduling lies in selecting appropriate buses to carry drones, considering their combined coverage capabilities with taxis, to achieve unpredictable emergency response performance. To address this issue, a data-driven and quantitative analysis approach was used to evaluate the combined emergency coverage capability of buses and taxis. First, a neural network model was used to learn daily taxi travel patterns and predict the number of unoccupied taxis in the urban spatiotemporal grid, as well as the travel time of taxis between urban grids. The predicted taxi travel patterns were combined with the drone emergency response process to guide the modeling of the individual drone-bus-taxi combined coverage capability, thereby guiding the optimization of the selection of buses carrying drones. The bus selection optimization problem was solved using a customized greedy algorithm for non-superimposed combined coverage gain.

[0159] In one embodiment of the present invention, in step 7, taxi demand and travel time are predicted. The prediction model predicts the number of unoccupied taxis in a grid at a fixed time, as well as the travel time required for a vehicle to travel from one grid to another.

[0160] Figure 2 The architecture for vehicle travel pattern prediction and bus recruitment is described. Urban grid data and temporal feature data are input into the MLP as context data for the two prediction tasks. Taxi trajectory data and context data from the first 5 days are used for model training, while data from the following 2 days are used for validation. Let M denote the layer number in the MLP, and H(k) and G(k) represent the outputs of prediction task 1 and prediction task 2 at the k-th layer, respectively. The inputs to the two MLP prediction tasks are H(0) and G(0), and H(M) and G(M) are their respective outputs. Specifically, the input H(0) is:

[0161] H(0) = [ξ, time, G] J (28)

[0162] Where, ξ j Let G represent the number of occupied taxis in the j-th sample, and let time and G indicate the time and grid ID of multiple unoccupied taxis, respectively. Similarly, the input G(0) is:

[0163]

[0164] in, This indicates that the taxi in the j-th sample is from G. O Arrive at G D The travel time. For example Figure 2 As shown, the inputs H(0) and G(0) of the two prediction tasks propagate along the black lines in the MLP. All layers in the MLP are fully connected layers (FC layers) with activation functions (AF). The outputs H(k) and G(k) of the k-th layer of the two prediction tasks are as follows:

[0165] H(k)=φ(H(k-1)w(k)+b(k))(30)

[0166] G(k)=φ(G(k-1)w(k)+b(k))(31)

[0167] Where w(k) are the weight parameters, b(k) are the bias terms, and φ(·) is the activation function. Leaky-Relu is used as the AF layer in the MLP to target potentially dead neurons during training. Therefore, the output of the activation function is:

[0168]

[0169] The Huber loss function is not only differentiable everywhere, but it also balances the accuracy and robustness of prediction results. Therefore, the Huber loss is used as the loss function for two MLP prediction tasks:

[0170]

[0171] Where δ is the decision threshold used to determine outliers, set to a default value of 1.0. y(k) and p(k) and p(k) represent the actual and predicted values ​​of the number of unoccupied taxis in the spatiotemporal grid, respectively. These are the actual and predicted travel times required for a taxi to travel from one spatiotemporal grid to another. During training, the losses for the two prediction tasks are calculated along their respective paths. Figure 2 Backpropagation is performed to correct the parameters in each layer. After training with two MLP prediction tasks, the number of unoccupied taxis in any spatiotemporal grid in the city and the travel time required for them to reach any grid can be predicted. The output of the prediction tasks is used to recruit buses equipped with drones.

[0172] In an embodiment of the invention, a joint coverage model of a single drone, bus, and taxi is established, where all bus trajectories are known, including grid ID sequences and timestamp sequences. Based on the bus emergency utility function, if the drone on bus b can respond to g... r In an emergency, otherwise, for Its combined coverage performance can be expressed as:

[0173]

[0174] UCU bi The larger the bus b i The greater the spatial and temporal coverage effect of drones, the more scientific it is than the traditional 0 or 1 coverage rate, because it comprehensively considers response delays, response duration, and emergency costs.

[0175] Considering a model for selecting buses equipped with drones for taxi relays, and given cost constraints, city managers can only select a subset of buses to equip with drones. Therefore, the objective is to maximize the selected bus set while keeping the total cost R within the budget B. The combined coverage effect:

[0176]

[0177] stR≤B(36).

[0178] In an embodiment of this invention, a non-overlapping coverage gain greedy algorithm is designed to select the optimal set of buses equipped with UAVs, simultaneously satisfying the rationality of emergency response decisions, budget feasibility, spatiotemporal grid coverage, coverage benefits, and computational efficiency. The basic idea of ​​NOCG-Greedy is to select a bus with the maximum joint coverage benefit, update the coverage benefit value of the spatiotemporal grid, and add the bus that contributes the most to the benefit value of the uncovered spatiotemporal grid to the currently selected bus set (while ensuring both spatiotemporal coverage and coverage benefits), until the budget is exceeded. Algorithm 1 is the NOCG-Greedy algorithm. In the allocation phase, it is necessary to control the number of selected buses to not exceed the budget limit of K. Therefore, the time complexity is O(K). In each outer loop, it is necessary to traverse all buses (O(m)) and all grids (O(n)) to calculate the coverage benefit of each bus. Therefore, the time complexity of the NOCG-Greedy algorithm is O(K·m·n). The space complexity is O(K+m), mainly used to store the covered spatiotemporal grid set and the selected bus set. In summary, NOCG-Greedy meets the computational efficiency requirements.

[0179]

[0180] The method of the present invention will be described below through a specific embodiment.

[0181] The proposed UBT scheme was evaluated using a real vehicle trajectory dataset to demonstrate its effectiveness.

[0182] (1) Setup: The experiment considered three important parameters that significantly affect the coverage performance of the drone: the number of buses carrying the drone, the proportion of candidate buses (number of candidate vehicles / total number of buses), and the ERD (the hovering time required for the drone at the emergency point). 5min Choose from [10min, 15min, 20min, 25min, 30min]. The number of buses equipped with drones is [10, 20, 30, 40]. 50 In the [15%, 20%, 25%, ...], the proportion of candidate buses is between [15%, 20%, 25%] 30％ The underlined value is the default value.

[0183] (2) Evaluation indicators: The following five indicators were used to evaluate the emergency coverage performance of UBT.

[0184] Emergency response delay, defined as the time elapsed from the occurrence of an emergency to the arrival of a drone at the emergency point, is critical to emergency response performance and should be minimized as much as possible.

[0185] Coverage area, defined as the maximum area that a drone can cover on a selected bus, should be as large as possible.

[0186] Spatiotemporal coverage is defined as the spatiotemporal coverage of a drone on a selected bus. It is affected by the Emergency Response Distance (ERD) of some emergency events. Specifically, if the ERD of some emergency events is too long, the drone may not be able to cover the area.

[0187] Vehicle recruitment costs: The UBT algorithm includes the recruitment costs of buses and taxis, while the UB algorithm (using only buses to carry drones) only includes the recruitment costs of buses.

[0188] Infrastructure costs, including drone and WPT costs.

[0189] (3) Baselines

[0190] Since UBT represents an initial exploration involving drones, buses, and taxis for urban emergency response, no existing baseline method is available. To evaluate the effectiveness of UBT, the UB algorithm was compared with three baseline methods for selecting drone-equipped buses; the method of this invention is abbreviated as UBT-TS.

[0191] 1) Random Selection (UBT-R). A specified number of buses are randomly selected from the candidate bus set.

[0192] 2) Maximum Time Greedy Algorithm (UBT-T). Selects a specified number of buses with the longest running time from the set of candidate buses.

[0193] 3) Maximum Space Greedy Algorithm (UBT-S). This algorithm selects a specified number of buses with the largest operating range from the candidate bus set. Note that UBT-R, UBT-T, and UBT-S all involve taxi relay cooperation.

[0194] 4) UB Algorithm. This algorithm uses only buses equipped with drones and selects a predetermined number of buses from a candidate bus set with the objective of maximizing spatiotemporal coverage. Note that the UB algorithm involves coordination between the drones and buses.

[0195] Unpredictable Emergency Response Performance Evaluation: Due to the strong spatiotemporal and random nature of unconventional emergency events, the spatiotemporal coverage of UAVs on selected buses was first analyzed to evaluate the emergency response performance of UBT. Furthermore, to evaluate the UBT's response performance to emergency events with different ERD requirements, 1200 emergency events located in the urban spatiotemporal grid were randomly generated, with the ERD of each emergency event in […]. 5min In the [10min, 15min, 20min, 25min, 30min], each ERD requirement includes emergency events at 200 different spatiotemporal grid locations. The emergency response efficiency of UBT is evaluated by calculating the spatiotemporal coverage of these emergency events.

[0196] (1) The impact of the number of candidate buses

[0197] Figure 3 The impact of the number of buses on the emergency response performance of various algorithms is shown. Figure 3 The spatial coverage of the UBT (Bus Transit Bus) scheme throughout the city was depicted as the number of candidate bus routes varied. Despite the increase in the number of candidate buses, the UBT scheme maintained full urban spatial coverage, with a coverage area consistently remaining at 2559 km². 2 The spatial coverage of the time-greedy, space-greedy, UB, and random algorithms is 2067km. 2 2543km 2 2505km 2 and 2343.5km 2The performance fluctuated. The UBT-T algorithm performed the worst among the five algorithms, dropping from 2172 square kilometers with a candidate bus ratio of 15% to 2039 square kilometers with a candidate bus ratio of 30%. This is because the UBT algorithm, while considering temporal and spatial features, introduces cooperation with relay taxis, further increasing the spatiotemporal coverage of the drone. In contrast, the maximum temporal and spatial greedy algorithms only consider one-dimensional features, leading to overlapping coverage of many urban spatial grids. While the UB algorithm considers temporal and spatial features, it does not include data-driven cooperation between buses and taxis, or between drones and taxis. This results in two situations where urban spatial grids cannot be covered: coverage failure due to the drone being too far from the bus to reach the emergency grid; and coverage failure due to the drone reaching the emergency grid but having insufficient remaining power to return to the original bus.

[0198] Figure 4 This shows the changes in spatiotemporal coverage. (Compared to...) Figure 3 Similarly, the UBT scheme maintains high spatiotemporal coverage, achieving 100% coverage regardless of the number of bus selections. The other four algorithms show significant performance degradation compared to the UBT-ST algorithm. This is because time-greedy and space-greedy algorithms only consider one-dimensional features, while spatiotemporal coverage reflects coverage performance across the spatiotemporal dimension. The UB algorithm focuses on the spatiotemporal dimension but ignores the relay benefits provided by taxis.

[0199] (2) The impact of the number of buses equipped with drones: Figure 5 The data shows how the emergency response performance of different algorithms changes with the number of buses equipped with drones. As the number of buses increases, the coverage area and spatiotemporal coverage of all five algorithms show an upward trend. Figure 5 As shown, the UBT-ST algorithm requires only 10 buses to cover an area of ​​2559 square kilometers in a city. In contrast, the UBT-T algorithm, with 50 buses carrying drones, only covers 2039 square kilometers. The coverage areas of other baselines are also smaller than UBT-ST. This is because UBT-TS comprehensively considers 2D features compared to UBT-S and UBT-T algorithms. Furthermore, the introduction of the NOCG-Greedy algorithm further reduces the overlapping coverage of spatiotemporal grids. Compared to the UB algorithm, the UBT-ST algorithm introduces cooperation with taxi relays, further expanding the emergency response range of drones. Figure 6This demonstrates the impact of the number of buses on urban emergency spatial-temporal coverage. Consistent with the aforementioned findings, the algorithm of this invention consistently exhibits the optimal urban emergency spatial-temporal coverage, requiring only 30 buses to achieve 100% coverage. In contrast, the UB scheme shows a significant slowdown in the improvement of urban emergency spatial-temporal coverage after the number of buses reaches 40, indicating that the UB scheme is approaching the critical point of diminishing marginal returns, and further increasing the number of buses will have a very limited effect on improving coverage.

[0200] (3) The impact of time period

[0201] Figure 7 The impact of time-varying time periods on the emergency response performance of the five algorithms is shown. For example... Figure 7 As shown, the coverage area of ​​all baselines fluctuates over time, while the coverage area of ​​the UBT-ST algorithm remains stable at 2559 square kilometers. This is because, compared to UBT-T, UBT-S, and UBT-Random, the UBT-ST algorithm considers not only 2D features but also collaboration with drones and the emergency response range of drones. The coverage area of ​​the UB algorithm at 0:00 at night is only 2023 square kilometers. This indicates that during the period from 0:00 to 4:00 at night, the public transportation routes operating in the city cannot meet the emergency response needs of the entire city. Figure 8 As shown, the UB algorithm has an emergency spatiotemporal coverage rate of only 72.74% at 0:00, while the UBT-ST algorithm has the highest emergency spatiotemporal coverage rate in all time periods, which demonstrates the superior urban emergency response performance of the UBT-ST algorithm.

[0202] (4) The impact of ERD

[0203] Figure 9 The results show a comparison of the coverage performance of five algorithms under 1200 random emergency events. Figure 9As shown, the UB algorithm has the shortest average emergency response time, at only 287.5 seconds, while the UBT-ST algorithm's drone emergency response time is 305 seconds, failing to achieve the shortest arrival time at the emergency point. The main reasons for this phenomenon can be attributed to the following two points: First, the UB algorithm only considers a single emergency mode where the drone takes off directly from the bus, and the selected emergency drone is based on the one closest to the emergency point, which limits its flexibility to some extent. In contrast, the algorithm of this invention introduces a relay taxi cooperation mechanism, which, although increasing the drone's emergency response delay to some extent, significantly improves coverage and robustness. Second, the statistical results are only based on random emergency events that the UB algorithm can cover, without considering the potential penalty for average response delay from emergency events it cannot cover, which may lead to an overestimation of the UB algorithm's performance. Furthermore, the UBT-Random algorithm has the longest drone emergency response delay, reaching 471 seconds, while the emergency response delays of the UBT-S and UBT-T algorithms are also much higher than those of the UB and UBT-ST algorithms. This is because the UB and UBT-TS algorithms comprehensively consider time and space dimensions during the bus selection process, thereby optimizing emergency response efficiency.

[0204] like Figure 10 As shown, the UBT-ST algorithm achieves an average drone hovering time of 2105 seconds, while the UB algorithm only achieves 1273 seconds. This advantage is mainly attributed to the UBT-ST algorithm's introduction of a relay taxi cooperation mechanism, which effectively reduces drone energy consumption, enabling drones to provide longer hovering times at emergency points. Other baseline algorithms also show significantly lower average hovering times than the UBT-ST algorithm. This is because the UBT-ST algorithm combines two-dimensional features (time and space) with the drone's emergency coverage range, further optimizing the collaborative coverage of buses, taxis, and drones, achieving minimized overlapping coverage.

[0205] Figure 11This paper demonstrates how the coverage of five algorithms changes with the Emergency Response Time (ERD) requirement of 1200 random emergency events. As the ERD requirement increases, the UBT-T algorithm consistently maintains the highest coverage, with an average coverage of 98.7%. In comparison, the average coverage of UBT-T, UBT-S, UBT-Random, and UB algorithms are 54.8%, 67.1%, 61%, and 69.6%, respectively. Notably, the UB algorithm exhibits the worst stability among the five algorithms; its coverage drops significantly from 45% to 25% when the ERD requirement increases from 25 minutes to 30 minutes. This phenomenon is mainly due to the UB algorithm's reliance on a single cooperative mode between buses and drones, resulting in significant drone power consumption on round-trip routes, thus failing to effectively cover emergency events with high ERD requirements. In contrast, the algorithm of this invention, by comprehensively considering two-dimensional features (time and space), introduces a relay taxi cooperative mechanism, significantly reducing the emergency power consumption of drones and further expanding their coverage range, thereby achieving higher coverage and stability.

[0206] (5) Emergency Cost Analysis

[0207] This invention provides a comprehensive cost assessment of five emergency response solutions, including vehicle recruitment costs and infrastructure costs (i.e., the acquisition costs of drones and WPT devices). The experiment first evaluated the daily vehicle recruitment costs required by the five algorithms under different spatiotemporal coverage rates in different cities ([80%, 85%, 90%, 95%]). Specifically, the study analyzed the daily costs required by each algorithm to achieve a specified spatiotemporal coverage rate. Subsequently, the infrastructure costs of the five algorithms were further evaluated. Notably, the vehicle recruitment costs for the UBT-ST, UBT-T, UBT-S, and UBT-Random algorithms include both daily bus recruitment costs and relay taxi recruitment costs, while the UB algorithm only includes daily bus recruitment costs.

[0208] Figure 12 The cost of the five algorithms was analyzed in detail. For example... Figure 12As shown, the number of buses required for all five algorithms increases with the increasing demand for emergency spatiotemporal coverage. The UBT-ST algorithm performs best, requiring only 5 buses to achieve 95% urban emergency spatiotemporal coverage. In contrast, the UBT-T, UBT-S, and UBT-Random algorithms require 430, 92, and 392 buses, respectively. The worst performing algorithm is UB, requiring 32, 40, and 138 buses for 80%-90% emergency spatiotemporal coverage, respectively. When the emergency spatiotemporal coverage reaches 95%, the UB algorithm requires as many as 5,000 buses, mainly due to the severe diminishing marginal returns of the UB algorithm.

[0209] Figure 13 The daily vehicle recruitment costs for five algorithms are shown. The UBT-ST algorithm has the lowest daily recruitment cost, averaging 29,000 RMB per day. In contrast, the UB algorithm has a daily recruitment cost as high as 781,500 RMB per day, while the daily recruitment costs of the UBT-T, UBT-S, and UBT-Random algorithms are also significantly higher than those of the UBT-ST algorithm. Figure 14 Further comparisons were made of the infrastructure costs of the five algorithms. The average infrastructure cost of the UBT-ST algorithm was RMB 281,200, while the infrastructure costs of UBT-T, UBT-S, UBT-Random, and UB algorithms were 1069.41%, 174.68%, 1003.95%, and 6827.45% higher than those of the UBT-ST algorithm, respectively. Specifically, when the urban emergency coverage rate was 95%, the infrastructure cost of the UB algorithm reached a staggering RMB 74.799 million, a cost that is unbearable for urban emergency response work. In conclusion, the UBT-ST algorithm demonstrates significant advantages in both daily vehicle recruitment costs and long-term infrastructure costs, making it more feasible and economical in practical applications.

[0210] This invention, based on large-scale real-world vehicle trajectory data, proposes for the first time a multimodal transport emergency response framework integrating drones, buses, and taxis. This framework aims to significantly improve the flexibility, economy, safety, and stability of urban emergency response without introducing new modes of transportation or consuming public resources. First, a vehicle movement model is constructed from the perspective of urban spatiotemporal grids. Based on this, a response process model covering four emergency scenarios is established, taking into account the energy consumption characteristics of drones. Subsequently, combined with a taxi recruitment model, a joint cost model is proposed that comprehensively considers emergency response delays, drone hovering time, and relay vehicle costs to optimize the emergency response performance of drones. Furthermore, a neural network model is used to mine the daily behavior patterns of taxis, further enhancing the model's practicality. Finally, from the perspective of urban emergency spatiotemporal coverage, a heuristic algorithm that balances spatiotemporal coverage overlay and coverage benefits is designed.

[0211] The proposed algorithm was quantitatively evaluated using a large-scale real-world vehicle trajectory dataset. Experimental results show that:

[0212] 1. Spatiotemporal coverage guidance: With only 30 drones, the method of this invention can achieve 100% spatiotemporal coverage of the city 24 hours a day, and the monitoring time of each spatiotemporal grid is not less than 5 minutes.

[0213] 2. Drone Energy Efficiency Oriented: With only 50 drones required, the method of this invention enables a single drone to monitor unpredictable urban emergencies for up to 35 minutes. Without significantly increasing emergency response delays, the maximum hovering time of the drones is improved by 65.4% compared to the baseline.

[0214] 3. Economic efficiency: With an infrastructure cost of only RMB 281,200, the method of this invention can achieve 95% emergency spatiotemporal coverage in cities, which is 6,827.45% lower than the baseline.

[0215] Next, with reference to the accompanying drawings, an emergency response dispatch system based on drones, buses, and taxis, according to an embodiment of the present invention, is described.

[0216] This emergency response and dispatch system, based on drones, buses, and taxis, includes:

[0217] The mobility modeling module is used to model the mobility of buses and taxis based on actual vehicle trajectory data to obtain vehicle mobility models.

[0218] The energy consumption model building module is used to calculate the energy consumption of UAVs during emergency response based on the vehicle movement model, and to analyze the emergency response mode of UAVs under various relay conditions from the perspective of energy consumption, and to build an energy consumption model for the emergency response process of a single UAV.

[0219] The relay cost calculation module is used to establish a taxi recruitment model and quantify the cost of relay taxis.

[0220] The joint cost establishment module is used to calculate the relay taxi cost based on the energy consumption model of the emergency response process of a single UAV and the taxi recruitment model. It comprehensively considers the UAV response delay, UAV response duration and relay taxi recruitment cost to establish a joint cost model for UAV emergency response.

[0221] The optimization module is used to optimize the joint cost model for UAV emergency response and select target UAVs to complete emergency tasks.

[0222] The prediction module is used to extend the emergency response scheduling of a single drone to the emergency response scheduling of multiple drones. It uses a taxi travel demand prediction model to predict taxi occupancy and travel time, and quantifies the coverage performance of taxis in the city.

[0223] The coverage module is used to calculate the spatiotemporal coverage of a single bus route based on the joint cost model of drone emergency response for individual buses and taxis, as well as the taxi travel demand prediction model.

[0224] The scheduling module is used to select the optimal set of buses equipped with drones from all bus routes based on the spatiotemporal coverage of a single bus route, using a greedy algorithm of non-overlapping joint coverage gain.

[0225] It should be noted that the foregoing explanation of the emergency response dispatch method embodiment based on drones, buses and taxis also applies to the emergency response dispatch system based on drones, buses and taxis in this embodiment, and will not be repeated here.

[0226] This invention proposes an emergency response dispatching method and system based on drones, buses, and taxis, outlining a UAV-ground transportation-based emergency response paradigm (UBT). This paradigm involves mounting drones on bus roofs for charging and inspection, and then coordinating with surrounding taxis to respond to unpredictable and random emergencies in cities. The invention establishes a multi-mode emergency response process model for a single drone considering the mobility of buses and the spatiotemporal characteristics of taxis. Combined with a data-driven spatiotemporal prediction model for taxis, a multi-drone emergency dispatching model is constructed, optimizing emergency response delays, drone hovering time, and relay costs, maximizing urban spatiotemporal coverage benefits. This invention provides an efficient and low-cost solution for urban emergency response, demonstrating significant application value.

[0227] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., refer to specific features, structures, materials, or characteristics described in connection with that embodiment or example, which are included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0228] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "N" means at least two, such as two, three, etc., unless otherwise explicitly specified.

[0229] Any process or method description in the flowchart or otherwise herein can be understood as representing a module, segment, or portion of code comprising one or N executable instructions for implementing custom logic functions or processes, and the scope of preferred embodiments of the invention includes additional implementations in which functions may be performed not in the order shown or discussed, including substantially simultaneously or in reverse order depending on the functions involved, as will be understood by those skilled in the art to which embodiments of the invention pertain.

Claims

1. An emergency response dispatch method based on drones, buses, and taxis, characterized in that, Includes the following steps: Step 1: Model the mobility of buses and taxis based on actual vehicle trajectory data to obtain vehicle mobility models; Step 2: Calculate the energy consumption of the UAV during the emergency response based on the vehicle movement model, and analyze the UAV emergency response mode under various relay conditions from the perspective of energy consumption, and construct an energy consumption model for the emergency response process of a single UAV. Step 3: Establish a taxi recruitment model to quantify the cost of relay taxis; Step 4: Based on the energy consumption model of the emergency response process of the individual UAV and the relay taxi cost calculated by the taxi recruitment model, and taking into account the UAV response delay, UAV response duration and relay taxi recruitment cost, establish a joint cost model for UAV emergency response. Step 5: Optimize the aforementioned joint cost model for UAV emergency response and select a target UAV to complete the emergency mission; Step 6: Extend the emergency response dispatch of a single drone to emergency response dispatch of multiple drones. Use the taxi travel demand prediction model to predict taxi occupancy and travel time, and quantify the coverage performance of taxis in the city. Step 7: Based on the joint cost model of drone emergency response for individual buses and taxis in Step 5 and the taxi travel demand prediction model in Step 6, calculate the spatiotemporal coverage of a single bus route. Step 8: Based on the spatiotemporal coverage of the single bus route, use the non-overlapping joint coverage gain greedy algorithm to select the optimal set of buses equipped with drones from all bus route sets, and determine the set of bus routes equipped with drones.

2. The method according to claim 1, characterized in that, Step 1 specifically includes: Discretize the entire city into a grid with the same side length. Given that the bus route and trajectory are known, the trajectory of any bus b is defined as follows: in, A tuple representing the grid ID and timestamp, i.e. (g k ,t k ), where k is the total number of trajectory points; The time when taxis participate in drone relay operations is divided into a series of time slots. Each time slot has the same duration. Only unoccupied taxis are considered for completing the drone relay mission, let N... un (g i ,s j ) indicates a time slot Period grid The number of unoccupied taxis for any timestamp Occurring in the grid In the event of an emergency, a binary variable is defined to indicate whether a relay taxi is carrying a drone from bus b. Wherein, r(g n ,t,v)=1 indicates that drones are allowed to operate from the grid containing the taxi grid. Unoccupied taxis were used to transport passengers to emergency grid areas.

3. The method according to claim 2, characterized in that, The takeoff of the drone from the bus to the emergency point is defined as the outbound journey, and the return journey is defined as the drone returning to the original bus after completing the mission. In step 2, the emergency response modes of drones under various relay conditions are analyzed from the perspective of energy consumption, including: There are four scenarios based on whether there are relay vehicles on the outbound and return journeys: Scenario 1: No relay vehicles on either the outbound or return journey; Scenario 2: Relay vehicles on the outbound journey but no relay vehicles on the return journey; Scenario 3: No relay vehicles on the outbound journey but relay vehicles on the return journey; Scenario 4: Relay vehicles on both the outbound and return journeys. drone trajectory Represented as: Among them, when drone u responds to emergency G at timestamp t0 p At that time, the flight trajectory was redefined G1(b) represents the grid where the bus is located at this timestamp, and the grid where the emergency point is located is G. p , express The encounter grid between the drone and the bus under the given conditions, where v is the relay taxi.

4. The method according to claim 3, characterized in that, In step 2, constructing the energy consumption model for the emergency response process of a single UAV includes: In scenario 1, the total energy consumption of the drone for: Wherein, α1 and α2 represent the power consumption of the UAV during flight and when hovering over the emergency point, respectively. For the hovering time of the drone at the emergency point, Dis(G1(b),G p () is a straight path. This represents the straight-line path from the emergency point to the bus grid after the drone completes its mission; Select the grid that provides the maximum response time for the drone as the final encounter grid between the drone and the bus. in, Let t be the average speed of the drone. meet The final encounter between the drone and the bus is represented by t0, which is the moment the emergency mission occurs. The time it takes for the drone to reach the emergency grid. E represents the remaining battery power of the drone before it performs an emergency mission, where λ is the percentage of the drone's battery capacity. c For drone battery capacity; In scenario 2, the total energy consumption of the drone for: Where ε is a default constant; The final encounter between the drone and the bus (grid) for: In scenario 3, the total energy consumption of the drone for: Unlike scenarios 1 and 2, the drone's hovering time In addition to energy consumption constraints, we also need to consider whether there are available taxis in the emergency point grid, and the final encounter grid between the drone and the bus. for: in, The travel time for a taxi from the emergency point grid to the meeting grid; Bus B arrives at the grid Timestamps, based on the expected bus trajectory Sure; Indicates drone hovering Afterwards, unoccupied taxis in the emergency response point grid can be used as relay vehicles; In scenario 4, the total energy consumption of the drone for: The taxi and the bus eventually met at the grid. for: in, This refers to the travel time for a taxi to travel from the emergency point grid to the meeting grid. Bus B arrives at the grid Timestamp.

5. The method according to claim 4, characterized in that, Step 3 specifically includes: Based on the routes and passenger occupancy information uploaded by taxis, calculate separately The cost of recruiting relay taxis for both outbound and return trips is rewarded based on the Euclidean distance between the taxi's departure and arrival grids. The taxi relay cost is calculated as follows: Among them, R taxi Taxi reward per unit length.

6. The method according to claim 5, characterized in that, Step 4 specifically includes: According to any bus UAV i Response delays under different drone emergency response modes Response duration and the cost of recruiting relay taxis Establish an emergency utility function for a single bus: in, for The joint emergency cost of getting off bus B; and The maximum values ​​for response delay, response duration, and relay taxi recruitment cost, respectively; ω i It is the weight corresponding to the cost; Establish a joint cost model for emergency response using drones, and determine the cost of bus b. i Ultimate emergency effectiveness The maximum emergency effectiveness under the four drone emergency response modes:

7. The method according to claim 6, characterized in that, Step 5 specifically includes: Taking into account the emergency response delays, duration, and costs associated with drones, the emergency task was assigned to bus b, which ultimately has the greatest emergency effectiveness. i Drones on the surface: in, Bus b at timestamp t0 n The number of drones.

8. The method according to claim 7, characterized in that, Step 6 specifically includes: A taxi travel demand prediction model is built using a neural network model to predict taxi passenger occupancy and the time required for a taxi to travel from one grid to another.

9. The method according to claim 8, characterized in that, Step 7 specifically includes: Based on the emergency utility function of a single bus, if the drone on bus b can respond to g... r In an emergency, otherwise, for The coverage model for a single bus is represented as follows: in, The larger the bus b i The greater the spatiotemporal coverage effect of drones; Considering cost constraints, select some buses to be equipped with drones. The goal is to maximize the selected bus set while keeping the total cost R within the budget B. The combined coverage effect: stR≤B in, Collect the selected buses The combined coverage effect.

10. An emergency response dispatch system based on drones, buses, and taxis, characterized in that, include: The mobility modeling module is used to model the mobility of buses and taxis based on actual vehicle trajectory data to obtain vehicle mobility models. The energy consumption model construction module is used to calculate the energy consumption of the UAV during the emergency process based on the vehicle movement model, and to analyze the UAV emergency response mode under various relay conditions from the perspective of energy consumption, and to construct an energy consumption model for the emergency response process of a single UAV. The relay cost calculation module is used to establish a taxi recruitment model and quantify the cost of relay taxis. The joint cost establishment module is used to establish a joint cost model for drone emergency response based on the energy consumption model of the emergency response process of the individual drone and the relay taxi cost calculated by the taxi recruitment model, taking into account the drone response delay, drone response duration and relay taxi recruitment cost. The optimization module is used to optimize the emergency joint cost model of the UAV and select the target UAV to complete the emergency mission; The prediction module is used to extend the emergency response scheduling of a single drone to the emergency response scheduling of multiple drones. It uses a taxi travel demand prediction model to predict taxi occupancy and travel time, and quantifies the coverage performance of taxis in the city. The coverage module is used to calculate the spatiotemporal coverage of a single bus route based on the joint cost model of drone emergency response for individual buses and taxis, as well as the taxi travel demand prediction model. The scheduling module is used to select the optimal set of buses equipped with drones from all bus route sets based on the spatiotemporal coverage of the single bus route using a greedy algorithm with non-overlapping joint coverage gain, thereby determining the set of bus routes equipped with drones.

Images