Oil-electric hybrid rescue material transportation following robot and control method

Abstract

The application discloses a kind of rescue material transport following robot based on oil-electric hybrid power and control method, to solve the problem of long-time heavy load, complex terrain, charging under the endurance of pure electric rescue material following robot, carrying, energy supplement problem.Its method obtains multi-source state, motion and environmental data, predicts power demand;Dynamically select pure electric, range, parallel or parking charging mode;Generate and motion power and motion instruction, drive robot following operation.The system includes hybrid power unit, following positioning, environmental perception module, main controller, energy management, motion control and driving execution unit.The present application effectively overcomes the endurance and energy supplement problem of pure electric robot, significantly improves the operation persistence, environmental adaptability, carrying efficiency and man-machine cooperation safety, optimizes system energy efficiency.

Description

Technical Field

[0001] This invention belongs to the field of fire-fighting equipment technology, specifically, it relates to a rescue material transportation follower robot based on hybrid electric power and its control method. Background Technology

[0002] In the field of modern emergency rescue, especially in the face of complex and ever-changing firefighting operations, the demand for efficient and reliable material transportation capabilities is increasingly urgent. With the rapid development of robotics technology and the continuous expansion of its application boundaries, intelligent transport robots have become an important technological direction for improving firefighting efficiency and ensuring the safety of rescue personnel. Their core value lies in their ability to replace manual labor in transporting heavy equipment (such as fire hoses, breathing apparatus, and demolition tools), especially in high-risk areas such as fire scenes, significantly reducing firefighters' physical exertion and exposure risks, thereby optimizing the overall rescue process.

[0003] In existing technologies, pure electric transport robots have seen initial applications in specific scenarios due to their advantages of zero emissions, low noise, and ease of integration. These robots typically use high-energy-density lithium-ion batteries as their sole power source and are driven by brushless DC motors. Correspondingly, to achieve human-robot collaborative movement, ultra-wideband (UWB) following technology, such as solutions based on dedicated chips like the DW1000, has been widely adopted. This technology, through precise time-difference ranging, achieves centimeter-level positioning accuracy and high refresh rate real-time tracking, enabling the robot to closely follow personnel and move flexibly in complex terrain or confined spaces, greatly improving the convenience and efficiency of operations. In terms of design principles, the combination of a pure electric solution and UWB following technology aims to build an autonomous mobile platform capable of intelligently sensing and responding to operator commands, with the goal of providing more targeted assistance for fire and rescue operations.

[0004] However, with the increasing complexity of fire and rescue missions and the more stringent requirements placed on the continuous operation capability of rescue equipment, the seemingly efficient combination of technologies has gradually revealed deep-seated inherent contradictions in its core power supply and operational strategies, making it increasingly limited in addressing new challenges. Specifically, traditional pure electric transport robots are constrained by the inherent energy density bottleneck and charging speed limitations of lithium batteries. Under actual heavy-load and long-term operating conditions, their endurance often fails to meet the demands, generally falling short of 2 hours. This seemingly simple problem of "insufficient endurance" cannot be effectively solved simply by increasing battery capacity; it involves a series of mutually restrictive technical challenges.

[0005] The reason for this is that configuring a large-capacity battery pack to extend battery life inevitably increases the robot's weight significantly. According to basic physics, increased robot weight first directly reduces its effective carrying efficiency; that is, with the same total load capacity, the weight of materials that can be transported decreases. Second, the heavier weight significantly increases the energy consumption required for motor operation, making the effect of simply increasing battery capacity to improve battery life greatly diminished, and potentially even leading to a diminishing returns problem. Furthermore, in fire rescue, especially in complex scenarios such as wildfires where there is no infrastructure, the convenience and timeliness of power replenishment are crucial. Pure electric solutions cannot quickly replenish power in these scenarios, and long charging cycles cause the robot to be idle for extended periods, severely impacting the continuity and timeliness of rescue operations. Moreover, attempting to use an external power cable for mobile charging not only greatly limits the robot's range of movement and flexibility but also poses a potential safety hazard in rugged or obstacle-filled fire scenes, even hindering the actions of rescue personnel. Therefore, the trade-off between range, carrying efficiency, and environmental adaptability, along with the strong dependence on external charging facilities, constitutes an unavoidable systemic bottleneck for existing pure electric transport robots when dealing with high-intensity, long-duration, and unreplenished fire rescue missions. Even if UWB following technology can provide precise navigation and coordination capabilities, if the underlying power system cannot provide continuous and sufficient energy support, the resulting efficiency improvement will be limited by the power supply bottleneck and cannot reach its maximum effectiveness, thus failing to truly meet the needs of long-term fire rescue.

[0006] Therefore, how to construct an intelligent transport robot power system that combines high endurance, strong carrying efficiency, excellent environmental adaptability and rapid energy replenishment, while effectively overcoming the inherent limitations of pure electric solutions in long-term operation and unstructured scenarios, has become a key challenge and a technical problem that needs to be solved by those skilled in the art. Summary of the Invention

[0007] The purpose of this invention is to provide, and primarily solve, [the problem / issue].

[0008] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0009] A control method for a rescue supplies transport following robot based on hybrid electric power includes:

[0010] S1, acquire the robot's real-time status data, which includes battery state of charge, fuel level, robot current speed, robot current acceleration, motor current, and robot posture data;

[0011] S2, acquire the position and motion state data of the target being followed, the position and motion state data being measured in real time by an ultra-wideband positioning module;

[0012] S3, acquire robot operating environment perception data, which is collected in real time by environmental sensors, including terrain slope information and obstacle distribution information;

[0013] S4 combines preset energy management strategies with real-time operating parameters to dynamically evaluate the output efficiency and remaining available energy of each power source;

[0014] S5. Dynamically select the robot's power working mode based on the battery state of charge, the fuel level, the output efficiency, and the remaining available energy.

[0015] S6, according to the optimal power operating mode, adjust the coordinated output of the electric system and the fuel power system;

[0016] S7, based on the location information of the target being followed, drives the rescue material transport follower robot to move autonomously and adjust its posture.

[0017] Furthermore, in this invention, acquiring the robot's real-time status data includes:

[0018] The battery management system module integrated inside the robot can acquire the total voltage, total current, voltage of each battery cell, real-time temperature of the battery pack, state of charge (SOC), and health status of the battery in real time. The battery management system module samples data through voltage and current sensors, estimates the SOC using a Kalman filter algorithm, and assesses the health status by combining historical charge and discharge data with internal resistance change trends.

[0019] The fuel power system monitoring module integrated on the robot can acquire real-time engine speed, fuel consumption rate, remaining fuel level in the fuel tank, engine operating temperature, lubricating oil pressure, and exhaust emissions. The remaining fuel level in the fuel tank is measured at multiple points using a liquid level sensor, and the fuel consumption rate is calculated based on injection pulse width, injection pressure, and engine speed parameters.

[0020] The robot's 3D attitude, angular velocity, angular acceleration, real-time 3D position coordinates, velocity, and acceleration are acquired in real time through the inertial measurement unit module and global positioning system module. The inertial measurement unit module provides the robot's 3D attitude, angular velocity, and angular acceleration, while the global positioning system module provides the robot's real-time 3D position coordinates, velocity, and acceleration. The sensor data is integrated and optimized using a sensor fusion algorithm based on extended Kalman filtering or unscented Kalman filtering.

[0021] The robot's current load status data can be obtained in real time by load sensor modules integrated on the robot chassis or suspension system.

[0022] Furthermore, in this invention, acquiring the position and motion state data of the target being followed includes:

[0023] By performing bidirectional ranging between the ultra-wideband tag set on the robot and the ultra-wideband anchor point set on the target, or by performing unidirectional or bidirectional ranging between the ultra-wideband tag set on the target and multiple ultra-wideband anchor points set on the robot, the three-dimensional relative position coordinates of the target relative to the robot are calculated in real time using the time difference of arrival or round-trip time measurement principle.

[0024] Time series analysis is performed on the three-dimensional relative position coordinates to calculate the real-time velocity and acceleration of the following target.

[0025] Furthermore, in this invention, acquiring robot operating environment perception data includes:

[0026] By scanning the robot's surroundings with LiDAR, a local 3D point cloud map is constructed to identify obstacles and extract their spatial distribution, relative position, and relative speed.

[0027] Environmental images are acquired using visual sensors to identify ground types, analyze terrain undulations, and estimate slopes.

[0028] The robot's absolute geographic coordinates are obtained through the Global Positioning System module, and combined with terrain elevation data, the slope of the terrain is determined.

[0029] Furthermore, in this invention, the specific process of step S4 is as follows:

[0030] S41, based on real-time battery state of charge, battery health status, battery voltage, current, and temperature data provided by the battery management system module, and combined with a preset battery internal resistance model and capacity decay model, accurately calculates the current available energy of the battery pack. The calculation formula is as follows:

[0031] E batt =SOH·C nom ·SOC·V avg

[0032] In the formula, E batt This indicates the current available energy of the battery pack, SOH indicates the battery's state of health, and C... nom The nominal capacity of the battery pack is indicated by V, and the state of charge (SOC) indicates the battery's state of charge. avg This indicates the average discharge voltage of the battery pack;

[0033] S42, based on the fuel remaining amount, real-time engine speed, fuel consumption rate and engine operating temperature provided by the fuel power system monitoring module, accurately calculates the remaining fuel in the fuel tank and converts it into energy that can be used for power generation.

[0034] S43 predicts the future performance degradation curve of the battery by analyzing the historical charge-discharge cycle count, depth, and long-term trend of the battery's health status for battery pack power systems; and predicts the engine's lifespan and potential failure risks by monitoring the engine's operating time, operating load, maintenance cycle, and wear indicators of key components for fuel power systems.

[0035] Furthermore, in this invention, step S5, the dynamic selection of the robot's power operating mode, includes the following steps:

[0036] S51, Define the energy management objective function:

[0037] Min{W1·C fuel +W2·C batt +W3·D life -W4·R range}

[0038] In the formula, W1, W2, W3, and W4 are the weighting factors for each item, and their values ​​are dynamically adjusted according to the current task type and priority; C fuel Fuel consumption per unit time; C batt D represents the amount of battery power consumed per unit time. life This refers to the lifespan loss of the battery and engine per unit time; R range Range per unit time;

[0039] S52, Construct the dynamic mode decision matrix: The decision matrix uses various possible dynamic working modes as rows and robot operating state parameters as columns; each cell in the matrix stores the "score" or "priority" for switching to or maintaining a certain dynamic mode under a specific combination of states;

[0040] S53, selects the optimal mode through decision-making logic and implements a mode switching strategy;

[0041] S54 performs real-time monitoring and feedback adjustments on the selected optimal power operating mode.

[0042] Furthermore, in this invention, step S6 is specifically defined as follows:

[0043] S61 performs precise control of the power system: based on the currently selected power operating mode, it sends specific instructions to the battery management system module and the motor controller.

[0044] S62 enables coordinated control and energy distribution between the electric system and the fuel power system: when the system is in hybrid mode, the total power demand is dynamically allocated between the battery and the fuel generator based on factors such as real-time power demand, battery state of charge, fuel reserve, and engine efficiency curve.

[0045] S63, Implementing a fault diagnosis and protection mechanism: Throughout the entire control process, the system continuously monitors all key parameters. When any parameter is detected to exceed the safety threshold or exhibit an abnormal trend, the fault diagnosis procedure will be initiated immediately.

[0046] Furthermore, in this invention, step S7 is specifically as follows:

[0047] S71 processes and filters the position information provided by the ultra-wideband positioning module: the received raw ultra-wideband position data needs to be smoothed by algorithms such as Kalman filtering, extended Kalman filtering or particle filtering to remove noise and improve the stability and continuity of position estimation; the filtered data will give the accurate real-time two-dimensional or three-dimensional coordinates of the target relative to the robot.

[0048] S72, Based on the filtered target position information, generate the robot's local motion target: Calculate the target relative position and target velocity that the robot should maintain based on the relative distance and direction between the robot and the target.

[0049] S73, perform local path planning and obstacle avoidance: while generating local moving targets, combine environmental perception data obtained by the robot's lidar, ultrasonic sensors or vision sensors to construct a local environment map around the robot.

[0050] S74, Implement motion control and attitude adjustment: Based on the local path planning results and the generated motion target, send precise control commands to the controllers of the drive motor and the steering mechanism;

[0051] S75 handles abnormal situations and ensures safety: During the following process, the system continuously monitors the signal strength and stability of the target being followed.

[0052] On the other hand, the present invention also provides a rescue supplies transport following robot based on hybrid electric power, comprising:

[0053] A hybrid power unit is used to provide driving electric power and manage fuel supply;

[0054] The tracking and positioning module is used to measure the position and motion status data of the target being tracked in real time.

[0055] An environmental perception module is used to acquire robot operating environment data, including terrain slope information and obstacle distribution information.

[0056] The main controller is used to acquire real-time status data of the robot, including battery state of charge, fuel level, robot current speed, robot current acceleration, motor current and robot posture data. It is also used to acquire the position and motion status data of the target being followed and the robot's working environment data acquired by the environmental perception module. Based on the acquired data, it predicts the robot's total power demand in the future time period.

[0057] The energy management module is used to dynamically select the robot's power operating mode based on the battery state of charge, the fuel level, and the predicted total power demand. The power operating mode includes pure electric mode, series range-extended mode, parallel auxiliary mode, and parking charging mode. Based on the selected power operating mode, the module outputs corresponding fuel generator control commands and / or motor drive commands.

[0058] The motion control module is used to generate motion control commands that meet the requirements of following accuracy and obstacle avoidance based on the position and motion state data of the target being followed, combined with the robot's own pose, robot working environment data and current power working mode.

[0059] The drive execution unit is used to receive and execute the fuel generator control commands and / or motor drive commands output by the energy management module and the motion control commands output by the motion control module, so as to drive the robot to perform follow-up operations.

[0060] Furthermore, in this invention, the hybrid power unit includes:

[0061] The fuel generator includes a battery pack, a fuel tank, and a power distribution and management circuit. The fuel generator is composed of an internal combustion engine coupled with a generator. The internal combustion engine is a small four-stroke gasoline engine, and the generator is a permanent magnet synchronous generator.

[0062] The battery pack is composed of high-energy-density lithium-ion battery cells connected in series and parallel, and has a high discharge rate characteristic.

[0063] The fuel tank is used to store fuel and is equipped with a fuel level sensor;

[0064] The power distribution and management circuit includes a DC-DC converter, an inverter, a charging module, and a power electronic switch, which are used to realize flexible switching and distribution of the electrical energy output by the fuel generator and the charging and discharging of the battery pack and the power supply of the motor.

[0065] Compared with the prior art, the present invention has the following beneficial effects:

[0066] (1) This invention effectively solves the problems of "insufficient range (generally <2 hours)" and "long charging cycle" caused by pure electric robots relying on a single battery for power by dynamically integrating the energy output of the battery pack and the fuel generator through a hybrid electric architecture. By dynamically selecting modes such as pure electric, series range extender, and parallel auxiliary, combined with the rapid fuel replenishment characteristics, the range under full fuel and full charge conditions is increased to ≥100km, which is more than 5 times that of the pure electric solution. Moreover, in the absence of charging facilities in the wild, power can be quickly restored by refueling, ensuring the continuity of rescue missions.

[0067] (2) This invention collects terrain slope and obstacle data in real time through an environmental perception module (LiDAR + visual sensor) and dynamically adjusts power output in conjunction with an energy management strategy. For example, it automatically switches to parallel mode when climbing or under high load, with the battery and fuel generator working together to output a peak power of 7kW, which improves traction compared to a pure electric solution; at the same time, it monitors the weight of materials in real time through a load sensor to optimize power distribution, effectively avoiding the contradiction that "large-capacity batteries increase weight and lead to a decrease in carrying efficiency", thus improving carrying efficiency.

[0068] (3) This invention employs "UWB positioning + multi-sensor fusion" following control technology, achieving centimeter-level positioning accuracy through a Kalman filter algorithm. The following distance control error is ≤ ±0.3m, ensuring that the robot closely follows rescue personnel in narrow spaces or areas with dense obstacles, avoiding the safety hazards of dragging cables or manual handling. Simultaneously, the energy management module optimizes power mode switching through an objective function, improving the overall system efficiency and effectively enhancing energy efficiency compared to a pure electric solution. Furthermore, fault diagnosis mechanisms (such as battery over-temperature protection and engine oil pressure monitoring) ensure operational safety and reduce equipment failure risks. Attached Figure Description

[0069] Figure 1 This is a flowchart of the robot control method of the present invention.

[0070] Figure 2 This is a system block diagram of the robot of the present invention. Detailed Implementation

[0071] The present invention will be further described below with reference to the accompanying drawings and embodiments. The embodiments of the present invention include, but are not limited to, the following embodiments.

[0072] Example

[0073] This invention discloses a rescue material transportation following robot and control method based on hybrid electric power. This technical solution constructs an intelligent transportation platform that takes into account long range, high load, rapid energy replenishment and excellent environmental adaptability through a sophisticated integrated high-efficiency hybrid electric range-extending power system and an advanced intelligent following control strategy, thereby substantially improving the material support capability and operational efficiency of fire rescue.

[0074] Specifically, such as Figure 2 As shown, this rescue supply transport robot is based on a modular intelligent control system, integrating seven key components: a hybrid power unit, a following and positioning module, an environmental perception module, a main controller, an energy management module, a motion control module, and a drive execution unit. Each module interacts with Ethernet in real time via a high-speed CAN bus, forming a closed-loop control system of "perception-decision-execution." It possesses autonomous following capabilities in complex environments, efficient power output, and reliable supply transport capabilities. Its overall design aims to meet the stringent requirements of rescue scenarios, achieving deep integration of power, perception, and control through multi-system collaborative operation, providing intelligent equipment support for emergency rescue, field operations, and other fields.

[0075] The hybrid power unit, serving as the robot's core power source, employs an intelligent collaborative power architecture and is integrated into a separate compartment in the middle of the robot. The compartment boasts an IP67 protection rating, with an external 3mm thick 5052 aluminum alloy armor plate and internal Nomex honeycomb flame-retardant cushioning material. It is equipped with dual redundant cooling fans and a temperature-sensing automatic start-stop system to ensure stable operation in environments ranging from -30℃ to 65℃. The unit's core function is to provide continuous driving power and achieve intelligent fuel-electric energy management, specifically consisting of a fuel generator system, a high-energy-density battery pack, and power distribution and management circuitry. The fuel generator system uses a single-cylinder four-stroke water-cooled gasoline engine, matched with a rare-earth permanent magnet synchronous generator, and integrates an intelligent electronic throttle and engine control unit. It supports high-efficiency fuel injection and lean-burn technology, and the power generation efficiency remains ≥90% in the 30%-100% load range. The battery pack is composed of 21700 ternary lithium-ion cells connected in series and parallel, and is equipped with a high-precision battery management system to realize real-time monitoring and protection of individual cell voltage, temperature, and current. The power distribution and management circuit realizes flexible energy switching between the fuel generator, battery pack, and drive motor through a DC / DC converter, bidirectional inverter, and intelligent power distribution unit, ensuring the continuity and efficiency of power output.

[0076] The following positioning module employs multi-source fusion positioning technology, utilizing multi-sensor data interaction combining visual recognition, UWB (ultra-wideband) navigation, and inertial navigation to achieve real-time tracking of the target's position and motion status. This module is equipped with three 16-line LiDARs, 16 ultrasonic sensors, two 2-megapixel global shutter cameras, and a 9-axis inertial measurement unit (IMU) to construct an all-around three-dimensional perception network. The UWB positioning system supports a hybrid TWR and TDOA positioning algorithm, achieving a static positioning accuracy of ±10cm and a dynamic positioning accuracy of ±30cm. The visual recognition unit incorporates a target detection algorithm, combining binocular parallax to calculate target distance, achieving a recognition accuracy of 98%. The IMU compensates for positioning drift by fusing Kalman filtering with visual / UWB data, ensuring target locking and trajectory prediction in complex environments. The following control strategy is based on the target's motion data from the past 200ms, using a second-order Kalman prediction model to estimate the future trajectory and adjust the robot's travel parameters in advance. The following distance control error is ±0.3m, and the angle error is ±3°, supporting multi-target switching and continuous anti-interference recognition.

[0077] The environmental perception module constructs a comprehensive three-dimensional perception network, collecting operational data such as terrain, obstacles, and environmental parameters through multi-sensor collaboration, providing support for path planning, obstacle avoidance control, and vehicle posture adjustment. A lidar array generates a real-time three-dimensional point cloud map of the environment, and a ground slope calculation algorithm obtains the terrain slope (accuracy ±1°) and road surface roughness. Ultrasonic sensors enable near-range obstacle detection, forming a three-level "far-medium-near" early warning mechanism. Temperature, humidity, and gas sensors monitor the safety of the operating environment in real time, triggering audible and visual alarms and reducing travel speed when combustible gas concentrations exceed limits. This data is transmitted to the main controller via a high-speed bus, providing the data foundation for the robot's ability to traverse complex terrains such as mountains, ruins, and mud.

[0078] The main controller, acting as the robot's "brain," employs a high-performance embedded computing platform responsible for data acquisition, decision analysis, and control command generation. Its core processor supports a wide operating temperature range of -40 to 85°C and is equipped with multi-interface communication modules to enable real-time data interaction with various sensors and actuators. On the software side, based on a Linux real-time kernel, it uses a federated Kalman filter algorithm to fuse state data such as battery SOC, fuel level, and motor current, combined with environmental perception of terrain slope and driving resistance, and employs a BP neural network model to predict the total power demand within the next 30 seconds, with a prediction error of <8%. The main controller's efficient data processing and decision-making capabilities ensure that the robot can quickly respond and optimize control strategies in dynamic environments.

[0079] The energy management module, based on a dynamic optimization control strategy, adaptively selects the optimal power operating mode according to battery state of charge, fuel level, predicted power demand, and operating scenario, achieving a balance between fuel efficiency and range. This module supports four modes: pure electric, series range extender, parallel auxiliary, and parking charging. Pure electric mode is suitable for low-power follow-up operations, with a range (unloaded) ≥15km; series range extender mode is activated during high power demand or low battery conditions, with a total system efficiency ≥25% and a range ≥100km on a full tank of fuel and a full charge; parallel auxiliary mode meets high-load scenarios such as climbing and acceleration, with a maximum output power of 7kW; parking charging mode automatically charges the battery when the robot is stationary, charging from 20% SOC to 80% in 1.5 hours. Mode switching is smoothly achieved through a fuzzy PID control algorithm, with a switching time <0.5s and output power fluctuation <5%, avoiding impact on the drive system.

[0080] The motion control module, based on model predictive control algorithms, integrates target trajectory tracking, environmental perception data, and vehicle attitude information to generate optimal motion control commands. Global path planning employs the A* algorithm, considering obstacle constraints to generate a smooth path; local obstacle avoidance control uses a dynamic window method, searching for the optimal control variable within the velocity space, achieving an obstacle avoidance response time of <0.3s and a safety distance of ≥0.5m. Vehicle attitude and speed control utilize PID algorithms to adjust the drive motor output torque, achieving speed control accuracy of ±0.2m / s and acceleration control accuracy of ±0.5m / s. 2 Combine IMU data to adjust the damping coefficient of the independent suspension system, control the vehicle body attitude change rate to <5° / s, and ensure that the levelness error of the material load platform is <3°.

[0081] The drive actuator, acting as the execution terminal for motion control, translates control commands into mechanical actions, driving the robot to achieve forward, backward, turning, and braking movements. The drive motor is a brushless DC motor with a rated power of 1.5kW / unit and a peak power of 2.5kW / unit, outputting a maximum torque of 120N·m through a planetary gear reducer. The steering system uses an Ackermann steering mechanism with a steering angle range of ±35° and a minimum turning radius of 1.5m. The braking system is a hydraulic disc brake with ABS anti-lock braking function, and a braking distance (speed 5m / s, load 500kg) <3m. The drive actuator monitors motor temperature, current, voltage, and other parameters in real time, triggering fault protection when an abnormality is detected to ensure operational safety.

[0082] like Figure 1 As shown, the control method of this robot is as follows:

[0083] The first step is to acquire real-time status data of the robot, comprehensively and accurately collecting key information about the rescue material transport robot and its external environment. This provides the necessary data foundation for subsequent energy management and power mode selection. Multimodal operational status data covers the robot's internal power system status, motion status, load status, and external positioning information related to the target being followed.

[0084] Data acquisition utilizes a battery management system module integrated within the robot to obtain real-time operating parameters of the power system. This includes the battery pack's total voltage, total current, voltage of each battery cell string, real-time temperature of the battery pack (e.g., temperature data from multiple points, such as the inside and surface of the battery module, obtained through a distributed temperature sensor array), state of charge (SOC), battery health (reflecting the degree of capacity degradation), and estimated remaining cycle life. The battery management system module samples data using precise voltage and current sensors and estimates the SOC using Kalman filtering or extended Kalman filtering algorithms, while simultaneously assessing the health status by combining historical charge / discharge data with internal resistance trends. All data is timestamped and uploaded to the main control unit's real-time data bus at a preset sampling frequency (e.g., every 50 milliseconds). The battery pack typically uses high-energy-density lithium-ion batteries with a typical nominal voltage of 48 volts and a capacity ranging from 100 to 200 amp-hours.

[0085] Secondly, the fuel power system monitoring module integrated into the robot acquires real-time operating parameters of the fuel power system. This includes real-time engine speed, fuel consumption rate, remaining fuel level in the fuel tank, engine operating temperature, lubricating oil pressure, and exhaust emissions. Remaining fuel level is measured non-contactly or contactlessly using a level sensor, and multi-point calibration is performed to improve accuracy. Fuel consumption rate is calculated based on parameters such as injection pulse width, injection pressure, and engine speed, or measured directly using a dedicated flow meter. Engine operating temperature and lubricating oil pressure are acquired using corresponding temperature and pressure sensors. All fuel power system data is timestamped and uploaded to the main control unit at the same sampling frequency as the electrical system data. The fuel power system typically consists of a small, high-efficiency internal combustion engine and a generator. The internal combustion engine displacement ranges from 50 to 200 ml, with a rated power between 2 kW and 5 kW, providing stable and efficient charging capabilities for the battery pack.

[0086] The robot's motion state data is acquired in real time through its inertial measurement unit (IMU) module and global positioning system (GPS) module. The IMU module provides the robot's three-dimensional attitude (roll, pitch, yaw), angular velocity, and angular acceleration. The GPS module provides the robot's real-time three-dimensional position coordinates, velocity, and acceleration. This sensor data is integrated and optimized using sensor fusion algorithms (e.g., based on extended Kalman filtering or unscented Kalman filtering) to eliminate single-sensor errors and improve the accuracy and robustness of positioning and attitude estimation. Motion state data is sampled at a high frequency (e.g., one hundred times per second) and synchronously transmitted to the main control unit. The GPS module typically supports multi-satellite reception to ensure centimeter-level positioning accuracy even in complex environments.

[0087] Simultaneously, load sensor modules integrated into the robot's chassis or suspension system acquire real-time data on the robot's current load status. These load sensor modules can be weighing sensors or force sensors, used to measure the real-time weight of materials carried on the robot platform. These sensors typically employ the strain gauge principle and, after precise calibration, provide high-precision weight measurements. Load data is crucial for assessing the robot's power requirements and stability. Load sensor data is uploaded to the main control unit at a preset frequency (e.g., ten times per second). The load sensor module's measurement range covers from zero kilograms to two hundred kilograms, with an accuracy of 0.1%.

[0088] Finally, the UWB (Ultra-Wideband) positioning module acquires the target's position information in real time. This module consists of an UWB tag mounted on the target and one or more UWB receivers mounted on the robot. The UWB tag periodically emits pulse signals at a high frequency (e.g., 50 times per second). Upon receiving these signals, the UWB receivers accurately calculate the target's relative position (distance and orientation) to the robot by measuring the time difference of arrival (TDOA), achieving centimeter-level positioning accuracy. The UWB module employs time-difference ranging (TDAR) and operates at a frequency range of 3.1 GHz to 10.6 GHz, with an effective ranging distance of up to 100 meters, providing high-precision positioning both indoors and outdoors. This position information is the core data for enabling the robot's following function.

[0089] During the acquisition of multimodal operational status data, all data is transmitted via an internal high-speed data bus (e.g., Controller Area Network bus or Ethernet), and the main control unit performs unified timestamp synchronization, data alignment, missing value imputation, and outlier filtering to ensure data consistency, integrity, and reliability. Data synchronization is calibrated using GPS time or a high-precision internal clock. Data alignment uses linear interpolation or spline interpolation methods to unify data from different sampling frequencies to a preset clock frequency. Missing value imputation is performed using historical data averaging, forward imputation, or machine learning prediction methods. Outlier filtering uses statistical methods, such as the three-standard-deviation rule, or machine learning-based anomaly detection algorithms to identify and remove outliers. These preprocessing steps aim to create a high-quality, unified time-series multidimensional dataset for subsequent complex analysis and decision-making.

[0090] The second step involves acquiring the position and motion state data of the target being followed. This data is measured in real time using an ultra-wideband positioning module. Based on the real-time acquired multimodal operational data, this step provides a deeper analysis of the robot's operating environment, current task stage, and physical load, transforming this information into structured data for energy management strategy decisions. The specific process is as follows:

[0091] First, the system analyzes the target location information provided by the ultra-wideband positioning module to determine if the robot is in follow mode. If the ultra-wideband positioning module continuously receives valid signals from the target, and the relative distance and relative speed between the robot and the target remain within a preset threshold range (e.g., relative distance between 1.5 meters and 5 meters, and relative speed difference less than 0.5 meters per second), the robot is determined to be in follow mode. In this mode, the system pays extra attention to the target's motion trend, including speed and rate of change of direction, to predict the robot's future motion needs. If the target signal is lost for more than a preset time (e.g., five seconds) or the relative distance exceeds a safe range (e.g., ten meters), the follow is considered interrupted, and the robot may enter autonomous navigation or standby mode.

[0092] Secondly, by combining data from the inertial measurement unit (IMU) module, the global positioning system (GPS) module, and possibly visual or lidar sensors, the complexity of the terrain in the robot's environment is perceived and assessed. For example, by analyzing the rate of change of the robot's pitch and roll angles, fluctuations in vertical acceleration, and instantaneous torque demands of the wheel motors, it can be identified whether the robot is climbing a slope, descending a slope, traversing rough terrain, or traveling on flat ground. The assessment of terrain complexity can be quantified into "low complexity," "medium complexity," and "high complexity" levels, each corresponding to different energy consumption models and power requirements. For example, continuous large pitch angle changes and high-frequency vibration amplitudes indicate highly complex terrain, requiring higher traction output and stronger power reserves.

[0093] Next, using the current load weight obtained from the load sensor module and combined with the robot's own weight, the total weight of the robot is calculated. Based on the total weight, the robot's current speed and acceleration, the instantaneous traction power required for the robot's current movement and the continuous power required to overcome friction and air resistance are estimated using a physical model or a pre-trained energy consumption model. The energy consumption model can be a multinomial regression model or a neural network model, with inputs including total weight, speed, acceleration, and terrain level, and the output being the instantaneous power demand. For example, when moving at a constant speed on flat ground, the main energy consumption is for overcoming rolling friction and the system's own operation; while climbing or accelerating, greater additional power is needed to overcome the increase in gravitational potential energy and kinetic energy. Load demand should be expressed in the form of real-time power (in watts).

[0094] Then, by combining historical task data, user-preset task parameters, and current battery and fuel levels, the system predicts the expected duration and total energy demand for the current task. For example, if the user sets a "long-distance transportation" task and the robot is carrying its maximum payload, the system will predict the approximate time required to complete the task based on the average energy consumption data of similar historical tasks, combined with the current terrain and speed. Simultaneously, the system will assess the remaining range or driving time of the battery and fuel under the current energy consumption mode and compare it with the predicted task duration to determine if there is an energy deficit. The prediction model can employ time-series prediction algorithms, such as Long Short-Term Memory networks or attention mechanism models, to improve prediction accuracy by learning the time dependencies and multivariate correlations in historical data.

[0095] Finally, all the above analysis results are comprehensively evaluated to generate a structured task scenario and load requirement report. This report includes, but is not limited to: current following status (e.g., "stable following", "following interrupted", "standby"), terrain level (e.g., "flat", "gentle slope", "rough"), real-time power requirement (e.g., 100 watts to 5 kilowatts), predicted task duration (e.g., two hours to eight hours), and energy supply gap warnings (e.g., "sufficient energy", "low battery warning", "low fuel warning"). This report is a key input for the next step of dynamically selecting the power operating mode, ensuring the comprehensiveness and accuracy of the decision.

[0096] The third step involves acquiring environmental perception data for the robot's operation. This data is collected in real-time by environmental sensors, including terrain slope information and obstacle distribution information. This step primarily involves using LiDAR to scan the robot's surroundings, constructing a local 3D point cloud map, identifying obstacles, and extracting their spatial distribution, relative position, and relative velocity. Visual sensors are used to acquire environmental images for ground type identification, terrain undulation analysis, and slope estimation. The robot's absolute geographic coordinates are obtained through a Global Positioning System (GPS) module, which, combined with terrain elevation data, helps determine the terrain slope.

[0097] The fourth step involves dynamically evaluating the output efficiency and remaining available energy of each power source by combining preset energy management strategies with real-time operating parameters. This step aims to conduct a comprehensive performance and energy assessment of the electric and fuel-powered systems of the rescue supplies transport robot, providing information for the subsequent power mode selection module to make decisions. The assessment results directly affect whether the robot can continuously and stably perform its tasks and maximize the service life of each power source.

[0098] Specifically, dynamically evaluating the output efficiency and remaining available energy of each power source includes the following sub-steps:

[0099] First, a refined assessment of the battery pack's remaining usable energy and output performance is conducted. Based on real-time battery state of charge, battery health status, battery voltage, current, and temperature data provided by the battery management system module, combined with preset battery internal resistance and capacity decay models, the current usable energy of the battery pack is accurately calculated. The calculation of usable battery energy needs to consider the impact of factors such as charge / discharge rate, temperature, health status, and internal impedance on battery capacity and maximum output power. For example, in low-temperature environments, the battery's internal resistance increases, leading to a decrease in its maximum discharge power and a corresponding reduction in usable energy. The assessment of output performance mainly focuses on the battery's maximum instantaneous discharge power, maximum continuous discharge power, and charging efficiency under the current temperature and health status. Simultaneously, by combining the predicted task duration and energy demand, the remaining driving range or driving time of the battery pack without charging is predicted.

[0100] For example, the available energy of a battery pack can be estimated using the following formula:

[0101] E batt =SOH·C nom ·SOC·V avg

[0102] In the formula, E batt The battery pack's current available energy is indicated by watt-hours (SOH), which represents the battery's state of health (0% to 100%). C nom The nominal capacity of the battery pack is indicated by ampere-hours (AH), and SOC indicates the state of charge of the battery (from 0% to 100%). V avg This represents the average discharge voltage of the battery pack (in volts). Both SOH and SOC are derived by the battery management system module through an online estimation algorithm, and are subject to temperature compensation and aging correction.

[0103] Secondly, the remaining usable energy and output efficiency of the fuel-powered system are assessed. Based on the fuel remaining amount, real-time engine speed, fuel consumption rate, and engine operating temperature provided by the fuel-powered system monitoring module, the remaining fuel in the fuel tank is accurately calculated and converted into energy that can be used for power generation. This calculation needs to consider the engine's instantaneous fuel consumption rate and load, generator efficiency, and fuel calorific value. For example, a fuel-powered generator has different fuel consumption curves under different loads. The system will query the corresponding fuel consumption rate based on the current expected power generation load to estimate the power generation time or total electrical energy that the remaining fuel can support. The output efficiency assessment mainly focuses on the fuel generator's maximum power generation, highest efficiency point, and response speed under the current operating temperature and load. Simultaneously, combined with the predicted task duration and energy demand, the remaining driving range or power generation time of the fuel-powered system without refueling is predicted. The fuel-powered system is mainly used to charge the battery pack; therefore, its output efficiency assessment should also include consideration of charging efficiency.

[0104] Next, the health status and degradation trends of both power sources are assessed. By analyzing the historical charge-discharge cycle count, depth, and long-term trends in health status of the battery pack, the future performance degradation curve of the battery is predicted. Similarly, for the internal combustion engine system, engine life and potential failure risks are predicted by monitoring engine operating time, operating load, maintenance cycle, and wear indicators of key components. These assessment results will be used to optimize the selection of power mode. For example, in specific situations, to extend battery life, the internal combustion engine system can be prioritized for power supply; or when the internal combustion engine system is about to require maintenance, it can be switched to a mode that relies primarily on battery operation.

[0105] Finally, the energy density, power density, energy conversion efficiency, and environmental adaptability of electric and gasoline-powered systems are comprehensively compared. For example, batteries offer high power response speed and high energy conversion efficiency, but their energy density is relatively low and they are significantly affected by temperature. Gasoline-powered systems have high energy density, long driving range, and wider adaptability to ambient temperatures, but their energy conversion efficiency is relatively low and they generate noise and emissions. This comprehensive comparison helps to dynamically weigh the advantages and disadvantages of each power source according to actual needs in different mission scenarios, thereby determining the power operating mode that best suits the current mission objectives. All evaluation results are output in a standardized form (e.g., percentage, watts, watt-hours) and provided to the next decision-making module.

[0106] The fifth step involves dynamically selecting the robot's power operating mode based on the battery's state of charge, fuel level, output efficiency, and remaining available energy. This step is the core of this control method, aiming to intelligently select the most suitable power operating mode based on the multimodal operating status data analysis results of the rescue material transport robot, the output efficiency and remaining available energy assessment of each power source, and preset energy management objectives. This dynamic selection process considers multiple constraints to achieve coordinated optimization of multiple objectives such as range, energy consumption, efficiency, safety, and lifespan. Figure 2 It provides a detailed explanation of the core principle framework for the dynamic selection of power operating modes.

[0107] Specifically, dynamically selecting the current optimal power operating mode includes the following sub-steps:

[0108] First, the energy management objective function is defined. This function is a multi-objective optimization function that comprehensively considers the following objectives: maximizing the robot's total range, minimizing total energy consumption (including electrical and fuel consumption), maximizing system efficiency, ensuring the lifespan of the battery pack and fuel-powered system, minimizing noise and emissions, and ensuring system operational safety. Each objective is assigned a weight factor, which can be dynamically adjusted based on the priority of the current task. For example, in emergency rescue missions, range and power output will have higher weights than energy consumption and lifespan.

[0109] The energy management objective function can be expressed as:

[0110] Min{W1·C fuel +W2·C batt +W3·D life -W4·R range}

[0111] In the formula, W1, W2, W3, and W4 are the weighting factors for each item, and their values ​​are dynamically adjusted according to the current task type and priority; C fuel Fuel consumption per unit time; C battD represents the amount of battery power consumed per unit time. life This refers to the lifespan loss of the battery and engine per unit time; R range The objective function defines the driving range per unit time. It aims to achieve optimal energy management by minimizing consumption and losses while maximizing driving range. All consumption and loss terms are standardized to eliminate dimensional differences and ensure the effectiveness of weight allocation.

[0112] Secondly, a power mode decision matrix is ​​constructed. This matrix uses various possible power operating modes as rows and various operating state parameters (e.g., battery state of charge, fuel remaining, current power demand, terrain complexity, ambient temperature, task priority, etc.) as columns. Each cell in the matrix stores a "score" or "priority" for switching to or maintaining a certain power mode under a specific combination of states. Possible power operating modes include:

[0113] First, pure electric mode: powered solely by a battery pack, the robot is driven by a motor. This mode is suitable for low-speed, low-load, short-distance following or indoor environments with strict requirements for noise and emissions.

[0114] Second, pure fuel mode: The fuel generator produces electricity, which directly drives the motor and may simultaneously charge the battery pack. This mode is suitable for scenarios requiring prolonged high power output and where the battery is low, but may be accompanied by noise and emissions.

[0115] Third, hybrid power mode (series hybrid): The fuel generator operates continuously, producing electricity, part of which directly powers the electric motor to drive the robot, and the other part charges the battery pack. This mode is suitable for medium to high loads and long-term operation, effectively utilizing the fuel generator's optimal efficiency range.

[0116] Fourth, hybrid mode (parallel hybrid): The battery pack and the fuel generator simultaneously power the motor to meet instantaneous high power demands. This mode is suitable for scenarios requiring explosive power output, such as climbing hills and acceleration.

[0117] Fifth, energy recovery mode: When the robot is going downhill or decelerating, the motor acts as a generator, converting kinetic energy into electrical energy to recharge the battery pack. This mode aims to improve energy utilization efficiency.

[0118] Sixth, Idle charging mode: When the robot stops moving, the fuel generator charges the battery pack to quickly restore power. This mode is suitable for when the robot is on standby or taking a short rest.

[0119] Secondly, the optimal mode is selected through decision-making logic. This decision-making logic can be a rule-based expert system, a fuzzy logic reasoning system, or a reinforcement learning model based on machine learning.

[0120] In rule-based expert systems:

[0121] If the battery state of charge is higher than a preset threshold (e.g., 70%) and the power demand is at a low to medium level, the pure electric mode is selected first to reduce fuel consumption and emissions.

[0122] If the battery's state of charge is below a preset threshold (e.g., 30%), and the task requires long-term operation or high power output, the system switches to hybrid mode (series or parallel) and determines whether to start the fuel generator to charge the battery or provide power directly, depending on the remaining fuel.

[0123] If the remaining fuel level falls below a safe threshold (e.g., 10%), the system will issue a warning and prioritize consuming battery power while advising the driver to refuel.

[0124] When the robot is going downhill or slowing down, regardless of the current primary power mode, it should prioritize entering the energy recovery mode to convert kinetic energy into electrical energy for storage.

[0125] When the robot remains stationary for an extended period and the battery's state of charge is low, the fuel generator automatically starts and enters idle charging mode until the battery's state of charge reaches a safe level.

[0126] In reinforcement learning-based models, the agent (decision-making module) learns, through interaction with the environment (robot and task scenario), which dynamic mode actions in different states will maximize long-term rewards (i.e., optimize the objective function). The observed states received by the agent include all multimodal operating state data and energy assessment results, the action space is the six dynamic operating modes mentioned above, and the reward function is designed based on the energy management objective function.

[0127] Next, implement the mode switching strategy. Mode switching not only requires selecting the optimal mode but also attention to the smoothness and safety of the switching process. For example, when switching from pure electric mode to hybrid mode, the fuel generator needs to go through processes such as startup, preheating, and acceleration. During this period, the battery pack needs to continue providing power to ensure uninterrupted robot operation. At the same time, the torque output of the motor and engine needs to be smoothly adjusted to avoid robot shaking or power interruption due to sudden changes. During the switching process, the robot's speed, acceleration, and vibration should be monitored in real time to ensure a smooth transition without affecting following stability.

[0128] Finally, the selected optimal power operating mode is monitored and adjusted in real time. After mode selection, the system continuously monitors the robot's actual operating performance, including actual energy consumption, speed, power output, and the response of each power source. If the actual operating effect deviates from expectations (e.g., energy consumption is higher than expected, or the battery state of charge drops too quickly), the system will re-evaluate and may reselect the mode. This closed-loop control mechanism ensures the dynamic adaptability of the power operating mode, enabling it to cope with various emergencies and environmental changes during operation. All decision-making and adjustment processes are recorded in the robot's control log for subsequent performance analysis and strategy optimization.

[0129] The sixth step is to adjust the coordinated output of the electric system and the fuel power system according to the optimal power working mode. This step is the specific execution link to realize the selected power working mode. It aims to coordinate the power output of the electric drive system and the fuel power system through precise control commands to meet the robot's current motion requirements and energy management goals.

[0130] Specifically, it includes the following sub-steps:

[0131] First, precise control is implemented for the power system. Based on the currently selected power operating mode (e.g., pure electric mode, hybrid mode), the main control unit sends specific instructions to the battery management system module and the motor controller.

[0132] In pure electric mode, the main control unit calculates the torque and speed that the motor should output based on the traction power required by the robot, and sends the corresponding pulse width modulation signal to the motor controller. Upon receiving the instruction, the motor controller precisely adjusts the current and voltage flowing to the drive motor to achieve the desired torque output and speed. Simultaneously, the battery management system module monitors the battery's discharge state, ensuring that the battery operates within safe voltage, current, and temperature ranges, avoiding overcharging, over-discharging, and overheating.

[0133] In hybrid mode, if the battery pack needs charging, the main control unit instructs the battery management system module to open the charging circuit and set the charging current and voltage so that the electrical energy generated by the fuel generator can efficiently and safely charge the battery. Meanwhile, the motor controller continues to adjust the motor output according to the robot's movement requirements.

[0134] In energy recovery mode, the main control unit instructs the motor controller to switch the motor to generator mode, adjust the feedback current, so that the motor converts kinetic energy into electrical energy when decelerating, and safely charges the battery pack through the battery management system module.

[0135] Secondly, precise control is applied to the fuel power system. Based on the currently selected power operating mode (e.g., pure fuel mode, hybrid mode, idle charging mode), the main control unit sends specific instructions to the engine controller and generator controller.

[0136] In pure fuel mode or hybrid mode, the main control unit calculates the required engine speed and load based on the portion of the robot's total power demand that is handled by the fuel power system. Upon receiving the command, the engine controller precisely adjusts the fuel injection quantity, ignition timing, intake air volume, and throttle opening to ensure the engine operates at the target speed and load, thereby outputting the corresponding mechanical power. The generator controller is responsible for converting the mechanical energy generated by the engine into electrical energy and, as needed, adjusting the generator's output voltage and current to directly supply the motor or charge the battery pack.

[0137] In idle charging mode, the main control unit instructs the engine controller to maintain the engine at the idle speed for optimal charging efficiency, and instructs the generator controller to charge the battery pack at a preset power until the battery state of charge reaches the preset value.

[0138] Engine control systems typically include closed-loop control, such as oxygen sensor feedback to control fuel injection, to optimize combustion efficiency and emissions. They also include sensors for temperature, pressure, and other parameters to monitor engine health; if an anomaly is detected, timely warnings and protective measures will be implemented.

[0139] Secondly, coordinated control and energy distribution between the electric system and the fuel-powered system are achieved. When the system is in hybrid mode, the main control unit acts as the core energy router. It dynamically allocates the total power demand between the battery and the fuel generator based on factors such as real-time power demand, battery state of charge, fuel reserve, and engine efficiency curves. For example, in scenarios with instantaneous high power demand, the battery can quickly respond to provide peak power, while the fuel generator provides continuous baseline power. This coordinated strategy is achieved through preset power distribution curves or model-based predictive control, ensuring a smooth transition between the two and avoiding power interruptions or overshoot. During the switching process, the torque compensation mechanism of the clutch or motor is used to ensure smooth power output and avoid mechanical shock.

[0140] Finally, a fault diagnosis and protection mechanism is implemented. Throughout the entire control process, the system continuously monitors all critical parameters (voltage, current, temperature, speed, pressure, etc.). If any parameter exceeds a safety threshold or exhibits an abnormal trend, the main control unit will immediately initiate a fault diagnosis procedure. For example, if the battery temperature is too high, the system will automatically reduce the charging or discharging current and activate the cooling fan; if the engine oil pressure is too low, the system will issue an alarm and may limit engine power output or even force a shutdown. These protection mechanisms aim to ensure the safe operation of the robot and its power system, preventing equipment damage and potential hazards. All fault information is recorded and uploaded to the remote monitoring platform via the communication module.

[0141] The seventh step involves driving the rescue supplies transport robot to move autonomously and adjust its posture based on the target's location information. This step is crucial for realizing the robot's core following function. It aims to utilize the precise location information of the target provided by the ultra-wideband positioning module, combined with the robot's motion state and environmental perception results, to plan the robot's motion trajectory and control its motion actuators, thereby achieving stable, safe, and efficient autonomous following and posture adjustment of the target.

[0142] Specifically, it includes the following sub-steps:

[0143] First, the position information provided by the ultra-wideband (UWB) positioning module is processed and filtered. While the UWB module provides high-precision relative position data, in complex environments, the signal may be affected by factors such as occlusion and multipath effects, leading to instantaneous data fluctuations. Therefore, the received raw UWB position data needs to be smoothed using algorithms such as Kalman filtering, extended Kalman filtering, or particle filtering to remove noise and improve the stability and continuity of the position estimation. The filtered data will provide the precise real-time two-dimensional or three-dimensional coordinates of the target relative to the robot (e.g., forward / backward distance, left / right offset, and vertical height difference).

[0144] Secondly, based on the filtered target position information, the robot's local motion targets are generated. The main control unit calculates the target-relative position and target velocity that the robot should maintain based on the relative distance and direction between the robot and the target. For example, the preset following distance can be between two and three meters, and the robot should always maintain the same orientation as the target. If the target moves forward, the robot needs to accelerate forward; if the target turns left, the robot needs to adjust its steering and rotate with the corresponding angular velocity. These local motion targets include the target linear velocity, target angular velocity, and target waypoint.

[0145] Next, local path planning and obstacle avoidance are performed. While generating the local moving target, the main control unit combines environmental perception data acquired by the robot's onboard LiDAR, ultrasonic sensors, or vision sensors to construct a local environment map around the robot. On this local map, a collision-free path from the robot's current position to the local moving target is planned. If obstacles exist on the path, a path planning algorithm (e.g., dynamic window method, artificial potential field method, or fast search tree algorithm) will generate the optimal path to bypass the obstacles. The obstacle avoidance strategy dynamically adjusts the robot's speed and steering based on the distance, size, and speed of the obstacles, ensuring that the robot can effectively avoid obstacles and return to the following path as quickly as possible, avoiding separation from the target. When the robot needs to avoid obstacles, it prioritizes ensuring that its relative position with the target is not significantly affected.

[0146] Then, motion control and attitude adjustment are implemented. The main control unit sends precise control commands to the controllers of the drive motors and the steering mechanism based on the local path planning results and the generated motion target (including linear velocity and angular velocity). The drive motor controllers adjust the speed and torque output of each wheel motor according to the required linear velocity and acceleration, enabling the robot to move forward, backward, and accelerate / decelerate. The steering mechanism controller adjusts the speed of each wheel using the steering motor or differential control according to the required angular velocity and steering angle, enabling the robot to turn and adjust its attitude. The control algorithm can be a proportional-integral-derivative controller, a fuzzy controller, or a model predictive controller. It compares the robot's current actual speed and attitude with the target value in real time and continuously adjusts the output commands to reduce errors. For example, by adjusting the speed difference between the left and right wheels, precise differential steering is achieved, ensuring that the robot's attitude remains consistent with the target being followed.

[0147] Finally, abnormal situation handling and safety assurance are implemented. During the following process, the system continuously monitors the signal strength and stability of the target. If the ultra-wideband signal is lost for an extended period (e.g., more than five seconds), or the distance between the robot and the target exceeds a safe range (e.g., ten meters), the system will determine that the following has been interrupted. At this time, the robot will automatically decelerate and stop, and issue an audible and visual alarm, awaiting manual intervention or signal recovery. Simultaneously, if the robot encounters an unavoidable obstacle or detects a potential collision risk during obstacle avoidance, the system will immediately execute an emergency stop operation. Furthermore, by fusing attitude data from the inertial measurement unit module, the system ensures that the robot maintains a stable attitude when navigating complex terrain, preventing rollover or loss of control. All abnormal situation handling prioritizes personnel safety and the integrity of materials. These safety mechanisms ensure that the robot can reliably perform following tasks in various complex environments.

[0148] The above embodiments are merely one of the preferred embodiments of the present invention and should not be used to limit the scope of protection of the present invention. Any modifications or refinements made to the main design concept and spirit of the present invention that are not of substantial significance, but solve the same technical problem as the present invention, should be included within the scope of protection of the present invention.

Claims

1. A control method for a rescue supplies transport following robot based on hybrid electric power, characterized in that, include: S1, acquire the robot's real-time status data, which includes battery state of charge, fuel level, robot current speed, robot current acceleration, motor current, and robot posture data; S2, acquire the position and motion state data of the target being followed, the position and motion state data being measured in real time by an ultra-wideband positioning module; S3, acquire robot operating environment perception data, which is collected in real time by environmental sensors, including terrain slope information and obstacle distribution information; S4, combining preset energy management strategies and real-time operating parameters, dynamically evaluates the output efficiency and remaining available energy of each power source; the specific process is as follows: S41, based on real-time battery state of charge, battery health status, battery voltage, current, and temperature data provided by the battery management system module, and combined with a preset battery internal resistance model and capacity decay model, accurately calculates the current available energy of the battery pack. The calculation formula is as follows: AND batt =SOH⋅C nom ⋅SOC⋅V avg In the formula, E batt This indicates the current available energy of the battery pack, SOH indicates the battery's state of health, and C... nom The nominal capacity of the battery pack is indicated by V, and the state of charge (SOC) indicates the battery's state of charge. avg This indicates the average discharge voltage of the battery pack; S42, based on the fuel remaining amount, real-time engine speed, fuel consumption rate and engine operating temperature provided by the fuel power system monitoring module, accurately calculates the remaining fuel in the fuel tank and converts it into energy that can be used for power generation. S43, for the battery pack power system, predicts the future performance degradation curve of the battery by analyzing the historical charge and discharge cycle count, depth, and long-term trend of the battery pack's health status. For fuel-powered systems, engine life and potential failure risks can be predicted by monitoring engine operating time, operating load, maintenance cycle, and wear indicators of key components. S5, dynamically select the robot's power operating mode based on the battery state of charge, the fuel level, the output efficiency, and the remaining available energy; wherein, dynamically selecting the robot's power operating mode includes the following steps: S51, Define the energy management objective function: Min{W1⋅C fuel +W2⋅C batt +W3⋅D life −W4⋅R range } In the formula, W1, W2, W3, and W4 are the weighting factors for each item, and their values ​​are dynamically adjusted according to the current task type and priority; C fuel Fuel consumption per unit time; C batt D represents the amount of battery power consumed per unit time. life This refers to the lifespan loss of the battery and engine per unit time; R range Range per unit time; S52, Construct the dynamic mode decision matrix: The decision matrix uses various possible dynamic working modes as rows and robot operating state parameters as columns; each cell in the matrix stores the "score" or "priority" for switching to or maintaining a certain dynamic mode under a specific combination of states; S53, selects the optimal mode through decision-making logic and implements a mode switching strategy; S54 performs real-time monitoring and feedback adjustments to the selected optimal power operating mode; S6, according to the aforementioned power operating mode, adjust the coordinated output of the electric system and the fuel power system; the specific steps are as follows: S61 performs precise control of the power system: based on the currently selected power operating mode, it sends specific instructions to the battery management system module and the motor controller. S62 enables coordinated control and energy distribution between the electric system and the fuel-powered system: when the system is in hybrid mode, it dynamically distributes the total power demand between the battery and the fuel generator based on real-time power demand, battery state of charge, fuel reserve, and engine efficiency curve. S63, Implementing a fault diagnosis and protection mechanism: Throughout the entire control process, the system continuously monitors all key parameters. When any parameter is detected to exceed the safety threshold or show an abnormal trend, the fault diagnosis procedure will be initiated immediately. S7, based on the location information of the target being followed, drives the rescue material transport follower robot to move autonomously and adjust its posture.

2. The control method for a rescue material transport following robot based on hybrid electric power as described in claim 1, characterized in that, The acquisition of the robot's real-time status data includes: The battery management system module integrated inside the robot can acquire the total voltage, total current, voltage of each battery cell, real-time temperature of the battery pack, state of charge (SOC), and health status of the battery in real time. The battery management system module samples data through voltage and current sensors, estimates the SOC using a Kalman filter algorithm, and assesses the health status by combining historical charge and discharge data with internal resistance change trends. The fuel power system monitoring module integrated on the robot can acquire real-time engine speed, fuel consumption rate, remaining fuel level in the fuel tank, engine operating temperature, lubricating oil pressure, and exhaust emissions. The remaining fuel level in the fuel tank is measured at multiple points using a liquid level sensor, and the fuel consumption rate is calculated based on injection pulse width, injection pressure, and engine speed parameters. The robot's 3D attitude, angular velocity, angular acceleration, real-time 3D position coordinates, velocity, and acceleration are acquired in real time through the inertial measurement unit module and global positioning system module. The inertial measurement unit module provides the robot's 3D attitude, angular velocity, and angular acceleration, while the global positioning system module provides the robot's real-time 3D position coordinates, velocity, and acceleration. The sensor data is integrated and optimized using a sensor fusion algorithm based on extended Kalman filtering or unscented Kalman filtering. The robot's current load status data can be obtained in real time by load sensor modules integrated on the robot chassis or suspension system.

3. The control method for a rescue material transport following robot based on hybrid electric power as described in claim 2, characterized in that, The acquisition of the position and motion state data of the target being followed includes: By performing bidirectional ranging between the ultra-wideband tag set on the robot and the ultra-wideband anchor point set on the target, or by performing unidirectional or bidirectional ranging between the ultra-wideband tag set on the target and multiple ultra-wideband anchor points set on the robot, the three-dimensional relative position coordinates of the target relative to the robot are calculated in real time using the time difference of arrival or round-trip time measurement principle. Time series analysis is performed on the three-dimensional relative position coordinates to calculate the real-time velocity and acceleration of the following target.

4. The control method for the rescue material transport following robot based on hybrid electric power as described in claim 3, characterized in that, The acquisition of robot operating environment perception data includes: By scanning the robot's surroundings with LiDAR, a local 3D point cloud map is constructed to identify obstacles and extract their spatial distribution, relative position, and relative speed. Environmental images are acquired using visual sensors to identify ground types, analyze terrain undulations, and estimate slopes. The robot's absolute geographic coordinates are obtained through the Global Positioning System module, and combined with terrain elevation data, the slope of the terrain is determined.

5. The control method for a rescue material transport following robot based on hybrid electric power as described in claim 4, characterized in that, The specific steps of step S7 are as follows: S71 processes and filters the position information provided by the ultra-wideband positioning module: the received raw ultra-wideband position data needs to be smoothed by Kalman filtering, extended Kalman filtering or particle filtering algorithms to remove noise and improve the stability and continuity of position estimation; the filtered data will give the accurate real-time two-dimensional or three-dimensional coordinates of the target relative to the robot. S72, Based on the filtered target position information, generate the robot's local motion target: Calculate the robot's target relative position and target velocity according to the relative distance and direction between the robot and the target; S73, perform local path planning and obstacle avoidance: while generating local moving targets, combine environmental perception data obtained by the robot's lidar, ultrasonic sensors or vision sensors to construct a local environment map around the robot. S74, Implement motion control and attitude adjustment: Based on the local path planning results and the generated motion target, send precise control commands to the controllers of the drive motor and the steering mechanism; S75 handles abnormal situations and ensures safety: During the following process, the system continuously monitors the signal strength and stability of the target being followed.

6. A rescue supplies transport following robot based on hybrid electric power, characterized in that, include: A hybrid power unit is used to provide driving electric power and manage fuel supply; The tracking and positioning module is used to measure the position and motion status data of the target being tracked in real time. An environmental perception module is used to acquire robot operating environment data, including terrain slope information and obstacle distribution information. The main controller is used to acquire real-time status data of the robot, including battery state of charge, fuel level, robot current speed, robot current acceleration, motor current and robot posture data. It is also used to acquire the position and motion status data of the target being followed and the robot's working environment data acquired by the environmental perception module. Based on the acquired data, it predicts the robot's total power demand in the future time period. The energy management module is used to dynamically select the robot's power operating mode based on the battery state of charge, the fuel level, and the total power demand. The power operating mode includes pure electric mode, series range-extended mode, parallel auxiliary mode, and parking charging mode. Based on the selected power operating mode, the module outputs corresponding fuel generator control commands and / or motor drive commands. The motion control module is used to generate motion control commands that meet the requirements of following accuracy and obstacle avoidance based on the position and motion state data of the target being followed, combined with the robot's own pose, robot working environment data and current power working mode. The drive execution unit is used to receive and execute the fuel generator control commands and / or motor drive commands output by the energy management module and the motion control commands output by the motion control module, so as to drive the robot to perform follow-up operations.

7. The rescue material transport following robot based on hybrid electric power as described in claim 6, characterized in that, The hybrid power unit includes: The fuel generator includes a battery pack, a fuel tank, and a power distribution and management circuit. The fuel generator is composed of an internal combustion engine coupled with a generator. The internal combustion engine is a small four-stroke gasoline engine, and the generator is a permanent magnet synchronous generator. The battery pack is composed of high-energy-density lithium-ion battery cells connected in series and parallel, and has a high discharge rate characteristic. The fuel tank is used to store fuel and is equipped with a fuel level sensor; The power distribution and management circuit includes a DC-DC converter, an inverter, a charging module, and a power electronic switch, which are used to realize flexible switching and distribution of the electrical energy output by the fuel generator and the charging and discharging of the battery pack and the power supply of the motor.

Images