Intelligent distribution system and method based on artificial intelligence and unmanned aerial vehicle

Abstract

The invention belongs to the technical field of intelligent logistics, and particularly relates to an intelligent distribution system and method based on artificial intelligence and an unmanned aerial vehicle. According to the invention, a three-dimensional intelligent distribution system with cooperation of a cloud decision center and edge calculation is constructed. The system dynamically divides a three-dimensional grid airspace through a space-time resource scheduling engine, and generates a flight path considering both efficiency and safety in combination with a multi-objective optimization algorithm; a standardized connection interface and a self-adaptive grabbing mechanism are adopted to achieve seamless connection of multiple goods, and a cloud-edge-end three-level disaster recovery system is constructed through a 5G-MEC network. According to the method, millimeter wave environment perception, model prediction control and reinforcement learning mechanisms are creatively fused, and full-process autonomous decision making and anomaly self-healing are achieved. According to the invention, the distribution efficiency, the equipment compatibility and the system reliability of urban unmanned aerial vehicle logistics are remarkably improved, and a core technical support is provided for constructing an air logistics network of a smart city.

Description

Technical Field

[0001] The present invention relates to the field of intelligent logistics technology, and specifically to an intelligent distribution system and method based on artificial intelligence and drones. Background Art

[0002] The current drone logistics sector faces two core bottlenecks: First, traditional distribution systems mostly use two-dimensional path planning models, which make it difficult to effectively coordinate the dynamically changing wind environment, temporary no-fly zones, and three-dimensional obstacles in urban airspace, resulting in poor route safety and low airspace utilization. Second, the hardware adaptability of logistics vehicles is insufficient. Most systems rely on customized cargo boxes and dedicated gripping devices. The existing electromagnetic locking mechanism is only suitable for cargo boxes of specific sizes and cannot meet the needs of multi-category mixing. In addition, when encountering communication interruptions, the mainstream solution relies on a preset return path and lacks the ability to make autonomous decisions based on environmental perception, which can easily lead to secondary accidents in complex urban areas. These technical shortcomings have seriously restricted the large-scale commercial application of drone logistics.

[0003] In response to the above-mentioned technical gaps, the present invention constructs a three-dimensional spatiotemporal resource scheduling model and an adaptive hardware system to break through the limitations of existing solutions in terms of airspace utilization efficiency, equipment compatibility, and system robustness, providing a new generation of technical infrastructure for smart logistics. Summary of the Invention

[0004] In order to solve the problems of the prior art, the present invention provides an intelligent distribution system and method based on artificial intelligence and drones.

[0005] In order to solve the above technical problems, the present invention is implemented through the following technical solutions: In the first aspect, an intelligent distribution system based on artificial intelligence and drones comprises:

[0006] User interaction terminal, merchant order receiving terminal, cloud decision center, drone cluster and cargo box adaptation module, including:

[0007] The cloud-based decision-making hub includes:

[0008] The spatiotemporal resource scheduling engine integrates user location data, merchant coordinates, and drone status parameters in real time to generate a three-dimensional distribution topology network;

[0009] Dynamic routing algorithm module, which uses a multi-objective optimization model to calculate the optimal flight path. The optimization objectives include shortest flight time, lowest energy consumption and maximum safety margin;

[0010] The cargo box adapter module is equipped with a standardized docking interface and a status monitoring unit, and is suitable for cargo boxes of any specifications;

[0011] The drone cluster configuration:

[0012] Heterogeneous navigation system, supporting redundant switching of GNSS, visual SLAM and UWB positioning;

[0013] Autonomous decision-making unit, which executes cloud commands while also having the ability to replan local paths;

[0014] Universal gripping mechanism, adapting to different box sizes through depth cameras and force sensors;

[0015] The various modules of the system achieve data collaboration through the 5G-MEC edge computing network, with a response delay of ≤50ms.

[0016] In the first aspect, we have built a technical framework for an intelligent distribution system based on artificial intelligence and drones, and achieved full-process automation and collaboration through modular design. The core value is reflected in three aspects:

[0017] Dynamic resource scheduling capabilities: The spatiotemporal resource scheduling engine and dynamic routing algorithm, combined with 3D grid modeling and multi-objective optimization, solve the path planning challenges in complex airspace environments, achieving a balance between delivery efficiency (time and energy consumption) and safety (obstacle avoidance and airspace control).

[0018] Heterogeneous device compatibility: The standardized interface and universal gripping mechanism design of the cargo box adapter module overcome the physical limitations of traditional drone vehicles through technologies such as electromagnetic intensity regulation and multi-spectral recognition, supporting seamless delivery of cargo of various specifications.

[0019] Autonomous Disaster Recovery System: The 5G-MEC edge computing network and offline reinforcement learning mechanisms create a three-level redundancy system: "cloud-edge-end." This ensures that essential services (such as route prediction and emergency return) can be maintained even in the event of communication interruptions, increasing system reliability to over 99.9%. This architecture, through the digitization of airspace resources, standardized hardware interfaces, and intelligent decision-making, provides a scalable technical foundation for urban logistics networks.

[0020] In a specific implementation of the first aspect, the method for constructing a three-dimensional distribution topology network includes:

[0021] A: Discretize the delivery airspace into a 50m×50m×20m grid;

[0022] B: Dynamically assign a travel cost value C = α·wind speed + β·obstacle density + γ·airspace control level to each grid, where α, β, and γ are real-time correction coefficients;

[0023] C: Searching for the optimal cost continuous grid sequence based on the improved Dijkstra algorithm;

[0024] Real-time correction coefficient generation rules:

[0025] α = 0.8 × (current wind speed / maximum allowable wind speed)^2, where the maximum allowable wind speed is set according to the drone model (15 m / s for DJI Matrice 300);

[0026] β = 1 / (1 + e^(-0.5 × obstacle density)), using the Sigmoid function for normalization;

[0027] γ = airspace control level × 0.3, where the control level is divided into 1-5 levels (level 5 is the highest limit).

[0028] In a specific embodiment of the first aspect, the universal gripping mechanism includes:

[0029] Retractable electromagnetic clamp, the magnetic field strength is automatically adjusted according to the box material, the adjustment range is 0.1-1.5 Tesla;

[0030] Multi-spectral scanning unit, which uses near-infrared spectroscopy to identify the box material and adjust the clamping strategy;

[0031] The six-dimensional force feedback system maintains a constant clamping force of 10N±0.5N during the grasping process.

[0032] In a specific implementation of the first aspect, the status monitoring unit performs:

[0033] Real-time detection of box dimensions through laser ranging with an accuracy of ±1cm;

[0034] Use inertial sensors to monitor the box's posture and trigger an alarm protocol when the tilt angle is greater than 15°;

[0035] Automatically identify the identification code on the box surface, analyze the cargo attributes and synchronize them to the cloud.

[0036] In a specific implementation of the first aspect, the system further includes an exception handling protocol:

[0037] When the drone encounters communication interruption, execute:

[0038] Step A: Switch to offline reinforcement learning mode and predict the recovery path based on historical trajectories;

[0039] Step B: If the offline mode lasts longer than 120 seconds, the autonomous return-to-home procedure is initiated, and the return path avoids the no-fly zone;

[0040] Step C: Upload the fault code to the nearest response node via the LoRa emergency channel.

[0041] In a specific implementation of the first aspect, the spatiotemporal resource scheduling engine of the cloud decision center specifically includes:

[0042] Airspace dynamic perception submodule, integrating weather radar data and ADS-B signals, with an update frequency of ≥10Hz;

[0043] The conflict prediction model calculates the probability of drone collision within the next 5 minutes based on Monte Carlo simulation, and triggers a route reset when the probability is greater than 0.1%;

[0044] The resource allocation optimizer uses mixed integer programming to solve the three-dimensional spatial grid occupancy plan with a solution time constraint of 500ms.

[0045] In a specific embodiment of the first aspect, the standardized docking interface includes:

[0046] Physical docking assembly: Magnetic buckle in accordance with ISO23387 standard, contact tolerance ±0.5mm;

[0047] Data communication protocol: Based on IEEE802.3 power carrier communication, synchronous transmission of cabinet ID code and status data;

[0048] Security verification mechanism: Use asymmetric encryption algorithm to verify the cabinet digital certificate, and the verification time is less than 100ms.

[0049] In a specific implementation of the first aspect, the 5G-MEC edge computing network implementation includes:

[0050] Edge servers deployed at drone take-off and landing points, with computing power ≥ 16 TFLOPS;

[0051] Data diversion strategy: key control instructions are directly connected to MEC nodes, and non-real-time monitoring data is transmitted back to the cloud;

[0052] Network redundancy design: When the main link delay is greater than 80ms, it automatically switches to the satellite communication backup link.

[0053] In the second aspect, an intelligent delivery method based on artificial intelligence and drones includes the following steps:

[0054] S1: Cabinet Registration

[0055] S1.1 The merchant terminal scans the box identification code and enters the cargo attributes and weight data;

[0056] The S1.2 cargo box adaptation module automatically calibrates the docking mechanism parameters to match the physical dimensions of the box;

[0057] S2: Cluster Scheduling

[0058] S2.1 The cloud decision center calculates the load balance of each drone L = current number of tasks / maximum carrying capacity;

[0059] S2.2 Select the UAV with the smallest L value to perform the new mission and generate a flight permit including the altitude layer assignment;

[0060] S3: Dynamic Delivery

[0061] The S3.1 drone uses millimeter-wave radar to scan the landing area and build a real-time point cloud map;

[0062] S3.2 uses the model predictive control (MPC) algorithm to adjust the flight attitude, and the landing positioning error is less than 5cm;

[0063] After S3.3 completes the cabinet handover, the battery life evaluation model is automatically updated. The evaluation formula is:

[0064] Remaining range R = (current power - safety threshold) / power consumption per unit distance × 0.9 redundancy coefficient.

[0065] In the second aspect, an operational paradigm for intelligent distribution methods was established, focusing on optimized control throughout the entire business lifecycle:

[0066] Process standardization: From automatic parameter matching for cabinet registration (±1cm precision calibration) to the load balancing algorithm for cluster scheduling (L value decision), a standardized operating process has been established to reduce efficiency losses caused by manual intervention.

[0067] Real-time dynamic optimization: The combination of millimeter-wave radar point cloud mapping and model predictive control (MPC) reduces landing positioning error to 5 cm, overcoming the environmental sensitivity bottleneck of traditional visual navigation. At the same time, a remaining range assessment model enables dynamic prediction of endurance (with a 0.9 redundancy factor).

[0068] Improved resource utilization: A combined altitude layer allocation strategy and airspace occupancy scheme using mixed integer programming have increased the drone swarm's delivery capacity per unit time by over 40%. This approach, through a data-driven closed-loop control mechanism, transforms the technical capabilities of the first aspect into quantifiable business metrics, forming a complete "perception-decision-execution" enhancement loop.

[0069] The beneficial effects of the present invention are:

[0070] 1. The present invention forms a two-tier decision-making architecture of "global optimization-local response" through the spatiotemporal resource scheduling of the cloud center and the real-time response of the edge computing nodes. In complex urban airspaces, it realizes: multi-dimensional path planning: integrating multiple constraints such as meteorological dynamics, obstacle distribution and airspace control to generate a three-dimensional navigation solution that takes into account both efficiency and safety, breaking through the limitations of traditional two-dimensional path planning; heterogeneous hardware collaboration: the combined design of standardized docking interfaces and adaptive gripping mechanisms enables drone clusters to have the ability to handle multiple specifications of goods, significantly expanding the applicable boundaries of logistics and distribution scenarios; system robustness enhancement: offline decision-making mechanisms based on reinforcement learning and multi-link communication redundancy ensure that basic service capabilities can be maintained in extreme environments or equipment failures, and build a resilient logistics network with self-healing capabilities;

[0071] 2. Systematic innovation, from technical architecture to operational methods, is driving a qualitative leap in logistics and distribution models: Intelligent business processes: Through the integration of technologies such as automatic parameter matching, millimeter-wave environmental perception, and model predictive control, traditional manual intervention is transformed into an autonomous decision-making process, significantly reducing operational complexity; Optimized resource scheduling: A spatiotemporal resource allocation model based on a three-dimensional airspace grid dynamically matches drone capacity with distribution needs, significantly improving logistics throughput per unit time; Scalable service capabilities: A modular design supports the rapid integration of new sensors and navigation equipment, enabling the system to continuously adapt to new distribution needs derived from the development of smart cities, laying the technical foundation for building an "integrated air-space-ground" logistics system. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Figure 1 It is a schematic diagram of the overall system architecture of the present invention.

[0073] Figure 2 It is a schematic diagram of the dynamic routing algorithm flow of the present invention.

[0074] Figure 3 It is a schematic diagram of the mixed integer programming optimization process of the present invention.

[0075] Figure 4 It is a schematic diagram of the exception handling protocol state of the present invention.

[0076] Figure 5 It is a schematic diagram of the cargo box adaptation process of the present invention. DETAILED DESCRIPTION

[0077] The following will be combined with the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0078] like Figures 1 to 5 An intelligent delivery system based on artificial intelligence and drones is shown.

[0079] 1. System hardware configuration and data parameters

[0080] 1. Cloud-based decision-making center

[0081] Spatiotemporal resource scheduling engine

[0082] Airspace dynamic perception submodule: Integrates X-band weather radar (operating frequency 9.3GHz, resolution 250m) and ADS-B receiver (1090MHz frequency band), real-time integration of urban three-dimensional map database (accuracy 0.5m), and data update frequency ≥10Hz.

[0083] Conflict prediction model: Monte Carlo simulation (10,000 times / cycle) is used to calculate the probability of drone collision in the next 5 minutes, with a trigger threshold of 0.1%.

[0084] Resource allocation optimizer: Configure the Gurobi9.5 solver to process mixed integer programming models (number of decision variables ≤ 5000) with a solution time constraint of 500ms.

[0085] 2. Drone swarms

[0086] Navigation system:

[0087] GNSS module: U-blox ZED-F9P (supports GPS / Galileo / Beidou, positioning accuracy 2cm RMS)

[0088] Visual SLAM: Intel RealSense D455 depth camera (TOF resolution 1280×720, 30fps)

[0089] UWB positioning: Decawave DW3000 chipset (6.5GHz frequency band, ranging error ±10cm)

[0090] Universal gripping mechanism:

[0091] Electromagnetic clamp: Magswitch2300N adjustable electromagnetic chuck (magnetic field strength 0.1-1.5T, response time <50ms)

[0092] Six-axis force sensor: ATIMini45 (range ±130N, resolution 0.01N)

[0093] Multispectral scanner: SpecimIQ (400-1000nm spectral range, recognition accuracy 99.8%)

[0094] 3. Cargo box adaptation module

[0095] Condition Monitoring Unit:

[0096] Laser ranging: SICKDT50 (measuring range 0.05-50m, accuracy ±1mm)

[0097] Inertial sensor: TDKICM-42688-P (three-axis accelerometer ±16g, gyroscope ±2000dps)

[0098] Identification code recognition: ZebraDS3608-UHD (supports QR / Datamatrix, decoding speed 300 times / second)

[0099] 4.5G-MEC edge computing network

[0100] Edge server: NVIDIA Jetson AGX Orin (computing power 275TOPS, 32GB RAM) Communication module: Huawei MH5000 5G module (downlink rate 2.5Gbps, latency <10ms)

[0101] Satellite backup link: Iridium9603N (transmission rate 176kbps, switching time <500ms)

[0102] 2. System Operation Process

[0103] Step 1: Cabinet Registration

[0104] Scan and input:

[0105] The merchant terminal uses ZebraDS3608 to scan the box identification code (GS1-128 standard) and enter the cargo attributes (size / weight / fragility level) into the cloud database.

[0106] The cargo box adapter module obtains the box size (length × width × height) through SICKDT50 laser ranging, with an error of ±1cm.

[0107] Parameter matching:

[0108] The docking mechanism automatically adjusts the spacing between the magnetic clips (stepper motor accuracy 0.01mm) to match the box size.

[0109] The preset magnetic field strength of the electromagnetic clamp is: 1.2T for metal box, 0.8T for composite material, and 0.3T for plastic.

[0110] Step 2: Cluster Scheduling

[0111] Load balancing calculation:

[0112] Cloud computing load balancing:

[0113] Select the drone with the smallest L value (such as DJI Matrice 300 RTK, with a maximum payload of 2.7 kg) and assign it a flight altitude layer (with an interval of 50 m).

[0114] Path planning:

[0115] The dynamic routing algorithm adopts NSGA-II multi-objective optimization with weight coefficients: time consumption 0.5, energy consumption 0.3, and security 0.2.

[0116] The output path is a three-dimensional coordinate sequence (WGS84 coordinate system, elevation accuracy ±0.1m).

[0117] Step 3: Dynamic Delivery

[0118] Landing Positioning:

[0119] Millimeter-wave radar: TIAWR2243 (77 GHz frequency band, point cloud density 1000 points / ㎡) builds a landing zone map.

[0120] MPC controller: Sampling period 50ms, prediction time domain 10 steps, control time domain 5 steps, achieving positioning error <3cm.

[0121] Cabinet handover:

[0122] The six-dimensional force sensor provides real-time feedback of the clamping force, and the PID controller maintains 10N±0.5N.

[0123] Data verification: SHA-256 encrypted box ID is transmitted to the drone via power carrier (HomePlugAV2 standard).

[0124] Battery life evaluation:

[0125] Remaining distance calculation:

[0126] Unit power consumption model: power consumption = 0.2 × load ratio + 0.05 × wind speed 2

[0127] (Load ratio = current load / maximum load)

[0128] 3. Exception handling mechanism

[0129] Communication interruption response

[0130] Offline path prediction:

[0131] A DQN reinforcement learning model (ε-greedy strategy, ε = 0.1) is used to generate a recovery path based on a historical trajectory library (over 10,000).

[0132] Local replanning frequency: 5 Hz.

[0133] Autonomous return:

[0134] The return path uses the A* algorithm to avoid no-fly zones (GIS geo-fence accuracy ±5m).

[0135] Emergency positioning: UWB beacon (Decawave DWM3000) provides local navigation.

[0136] Fault reporting:

[0137] LoRa module: Semtech SX1276 (868 MHz frequency band, transmission distance 10 km), sending fault code (16-bit CRC check).

[0138] Grid size determination method: Based on the urban building height distribution characteristics (90% of buildings are ≤100m), a tiered airspace management strategy is adopted. Computational fluid dynamics (CFD) simulations show that when the grid height is ≥20m, the standard deviation of wind field disturbance is less than 2m / s (test data see the table below), meeting the drone's attitude stability threshold.

[0139] Grid height (m) Wind speed standard deviation (m / s) Turbulence intensity (%) 10 3.2 15.8 20 1.7 8.9 30 1.5 7.2

[0140] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent distribution system based on artificial intelligence and drones, characterized by: include: User interaction terminal, merchant order receiving terminal, cloud decision center, drone cluster and cargo box adaptation module, including: The cloud-based decision-making hub includes: The spatiotemporal resource scheduling engine integrates user location data, merchant coordinates, and drone status parameters in real time to generate a three-dimensional distribution topology network; Dynamic routing algorithm module, which uses a multi-objective optimization model to calculate the optimal flight path. The optimization objectives include shortest flight time, lowest energy consumption and maximum safety margin; The cargo box adapter module is equipped with a standardized docking interface and a status monitoring unit, and is suitable for cargo boxes of any specifications; The drone cluster configuration: Heterogeneous navigation system, supporting redundant switching of GNSS, visual SLAM and UWB positioning; Autonomous decision-making unit, which executes cloud commands while also having the ability to replan local paths; Universal gripping mechanism, adapting to different box sizes through depth cameras and force sensors; The various modules of the system achieve data collaboration through the 5G-MEC edge computing network, with a response delay of ≤50ms.

2. The intelligent distribution system and method based on artificial intelligence and drones according to claim 1, characterized in that: The three-dimensional distribution topology network construction method includes: A: Discretize the delivery airspace into a 50m×50m×20m grid; B: Dynamically assign a travel cost value C = α·wind speed + β·obstacle density + γ·airspace control level to each grid, where α, β, and γ are real-time correction coefficients; C: Searching for the optimal cost continuous grid sequence based on the improved Dijkstra algorithm; Real-time correction coefficient generation rules: α = 0.8 × (current wind speed / maximum allowable wind speed)^2, where the maximum allowable wind speed is set according to the drone model (15 m / s for DJI Matrice 300); β = 1 / (1 + e^(-0.5 × obstacle density)), using the Sigmoid function for normalization; γ = airspace control level × 0.3, where the control level is divided into 1-5 levels (level 5 is the highest limit).

3. The intelligent distribution system and method based on artificial intelligence and drones according to claim 1, characterized in that: The universal gripping mechanism comprises: Retractable electromagnetic clamp, the magnetic field strength is automatically adjusted according to the box material, the adjustment range is 0.1-1.5 Tesla; Multi-spectral scanning unit, which uses near-infrared spectroscopy to identify the box material and adjust the clamping strategy; The six-dimensional force feedback system maintains a constant clamping force of 10N±0.5N during the grasping process.

4. The intelligent distribution system and method based on artificial intelligence and drones according to claim 1, characterized in that: The condition monitoring unit performs: Real-time detection of box dimensions through laser ranging with an accuracy of ±1cm; Use inertial sensors to monitor the box's posture and trigger an alarm protocol when the tilt angle is greater than 15°; Automatically identify the identification code on the box surface, analyze the cargo attributes and synchronize them to the cloud.

5. The intelligent distribution system and method based on artificial intelligence and drones according to claim 1, characterized in that: The system also includes an exception handling protocol: When the drone encounters communication interruption, execute: Step A: Switch to offline reinforcement learning mode and predict the recovery path based on historical trajectories; Step B: If the offline mode lasts longer than 120 seconds, the autonomous return-to-home procedure is initiated, and the return path avoids the no-fly zone; Step C: Upload the fault code to the nearest response node via the LoRa emergency channel.

6. The intelligent distribution system and method based on artificial intelligence and drones according to claim 1, characterized in that: The spatiotemporal resource scheduling engine of the cloud decision-making center specifically includes: Airspace dynamic perception submodule, integrating weather radar data and ADS-B signals, with an update frequency of ≥10Hz; The conflict prediction model calculates the probability of drone collision within the next 5 minutes based on Monte Carlo simulation, and triggers a route reset when the probability is greater than 0.1%; The resource allocation optimizer uses mixed integer programming to solve the three-dimensional spatial grid occupancy plan with a solution time constraint of 500ms.

7. The intelligent distribution system and method based on artificial intelligence and drones according to claim 1, characterized in that: The standardized connection interface includes: Physical docking assembly: Magnetic buckle in accordance with ISO23387 standard, contact tolerance ±0.5mm; Data communication protocol: Based on IEEE802.3 power carrier communication, synchronous transmission of cabinet ID code and status data; Security verification mechanism: Use asymmetric encryption algorithm to verify the cabinet digital certificate, and the verification time is less than 100ms.

8. The intelligent distribution system and method based on artificial intelligence and drones according to claim 1, characterized in that: The 5G-MEC edge computing network implementation includes: Edge servers deployed at drone take-off and landing points, with computing power ≥ 16 TFLOPS; Data diversion strategy: key control instructions are directly connected to MEC nodes, and non-real-time monitoring data is transmitted back to the cloud; Network redundancy design: When the main link delay is greater than 80ms, it automatically switches to the satellite communication backup link.

9. An intelligent distribution method based on artificial intelligence and drones, characterized in that: The following steps are involved: S1: Cabinet Registration S1.1 The merchant terminal scans the box identification code and enters the cargo attributes and weight data; The S1.2 cargo box adaptation module automatically calibrates the docking mechanism parameters to match the physical dimensions of the box; S2: Cluster Scheduling S2.1 The cloud decision center calculates the load balance of each drone L = current number of tasks / maximum carrying capacity; S2.2 Select the UAV with the smallest L value to perform the new mission and generate a flight permit including the altitude layer assignment; S3: Dynamic Delivery The S3.1 drone uses millimeter-wave radar to scan the landing area and build a real-time point cloud map; S3.2 uses the model predictive control (MPC) algorithm to adjust the flight attitude, and the landing positioning error is less than 5cm; After S3.3 completes the cabinet handover, the battery life evaluation model is automatically updated. The evaluation formula is: Remaining range R = (current power - safety threshold) / power consumption per unit distance × 0.9 redundancy coefficient.

Images