A self-balancing vehicle multi-vehicle platoon control system based on attitude safety constraints

Abstract

The present application relates to the technical field of intelligent transportation and vehicle control, and discloses a multi-vehicle platoon control system for self-balancing vehicles based on attitude safety constraints, which comprises multiple cooperative self-balancing vehicles, a vehicle-mounted main control unit calculates an attitude risk index based on sensing data, adjusts platoon stiffness in real time through an impedance control strategy, and actively reduces the stiffness to obtain attitude margin when the risk increases; when the actuator tends to be saturated, the system generates a virtual vehicle length parameter containing a momentum unloading buffer distance and broadcasts it, and uses an asymmetric inflation strategy to induce adjacent vehicles to give up a longitudinal operation window, so that the vehicle executes a pulse maneuver to reset the control moment gyro angular momentum. Through risk adaptive variable stiffness and cooperative space expansion strategy, the present application effectively solves the conflict between following precision and attitude stability in the platoon, realizes online recovery of the actuator capacity, and improves the safety of self-balancing vehicle cluster driving.

Description

Technical Field

[0001] This invention relates to the field of intelligent transportation and vehicle control technology, specifically to a self-balancing multi-vehicle platooning control system based on attitude safety constraints. Background Technology

[0002] Self-balancing vehicles (such as electric unicycles and two-wheeled self-balancing vehicles) have been widely used in short-distance commuting, logistics delivery, and park patrols due to their compact structure, maneuverability, and small footprint. With the development of intelligent connected vehicle technology, forming platoons of multiple self-balancing vehicles for coordinated driving can effectively reduce wind resistance and energy consumption, improve road traffic efficiency, and reduce human supervision costs, making it an important research direction in this field.

[0003] However, existing multi-vehicle platooning control technologies are mostly developed based on statically stable platforms such as four-wheeled vehicles. Their control logic typically assumes that the vehicle chassis has absolute attitude stability and focuses on minimizing vehicle spacing errors through high-gain feedback to achieve tight following. This traditional control strategy has limitations when directly applied to self-balancing vehicle platooning. Self-balancing vehicles are typical underactuated nonlinear systems, and their longitudinal motion is strongly coupled with their vehicle attitude; that is, vehicle acceleration or deceleration must be achieved by changing the vehicle's pitch angle. During platooning, if the speed of the preceding vehicle fluctuates drastically, the controller of the following vehicle often requests large accelerations or decelerations to maintain the intended high-precision following, forcing the vehicle to tilt significantly.

[0004] This forced correction of following errors can easily cause the vehicle's actuators (such as hub motors or control torque gyroscopes for balancing) to quickly reach their physical limits, i.e., enter a saturation state. Once the actuators are saturated, the vehicle will lose the torque margin required to adjust its attitude, leading to loss of attitude control or even a crash. Furthermore, in tight platooning formations, the front and rear space of vehicles is strictly locked. When a vehicle detects that its actuators are about to saturate and needs to perform an attitude recovery operation (such as unloading momentum through specific acceleration and deceleration maneuvers), it cannot obtain sufficient longitudinal operating space due to the existing rigid vehicle spacing maintenance logic. This contradiction between physical constraints and control logic makes self-balancing vehicle platoons vulnerable to disturbances, hindering the practical application of this type of vehicle swarm technology. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a self-balancing multi-vehicle platooning control system based on attitude safety constraints. This solves the problems of existing self-balancing vehicle platooning control strategies failing to fully consider vehicle attitude stability boundaries, leading to actuator saturation and instability under strong maneuvering or disturbance conditions due to the pursuit of following accuracy, and the difficulty for a single vehicle to recover its attitude control capability through active maneuvering under the constraints of a compact platooning space.

[0006] To achieve the above objectives, the present invention provides a self-balancing multi-vehicle platooning control system based on attitude safety constraints.

[0007] The system aims to address the problem of self-balancing vehicles becoming unstable due to the saturation of their attitude control capabilities while pursuing close following during platooning. It achieves a dynamic balance between vehicle attitude stability and platooning following ability through a multi-level risk adaptive control strategy.

[0008] The system comprises multiple self-balancing vehicles working collaboratively, each equipped with an execution system, a sensing system, a communication system, and an onboard main control unit. The execution system integrates a hub motor module and a control torque gyroscope module. The hub motors provide longitudinal drive torque, while the control torque gyroscope module generates high-frequency precession gyro torque based on the principle of conservation of angular momentum to maintain vehicle balance. The sensing system uses an inertial measurement unit, wheel speed sensors, and a state encoder within the control torque gyroscope module to collect real-time data on the vehicle's three-axis attitude, drive wheel speeds, and the gyroscope frame's deflection angle and rotor speed.

[0009] The onboard main control unit is equipped with state monitoring, risk assessment, variable stiffness control, and collaborative unloading decision-making logic, which achieves attitude safety control of the formation through the following steps: First, the system establishes a physical characteristics-based attitude risk assessment mechanism. The risk assessment module monitors the operating status of the control moment gyroscope module in real time and calculates the saturation index, which represents the ratio of the current frame angle to the maximum mechanical deflection angle. The system constructs a three-dimensional attitude safety envelope that includes the vehicle pitch angle, pitch rate, and saturation index, dividing the vehicle's operating state into an absolute safety zone, a risk warning zone, and an instability critical zone. Based on the vehicle's position within these zones, the attitude risk index is calculated; within the risk warning zone, the index increases linearly with saturation; once the instability critical zone is entered, the index increases exponentially to quickly respond to potential instability risks.

[0010] Secondly, the system employs a risk-adaptive variable stiffness impedance control strategy. The controller models the longitudinal following relationship between the vehicle and the preceding vehicle as a virtual impedance system containing virtual springs and virtual dampers. The onboard main control unit adjusts the stiffness coefficient of the virtual spring based on the real-time calculated attitude risk index, with the two being negatively correlated. When the attitude risk is low, the nominal stiffness is maintained to ensure following accuracy; when the attitude risk increases, the stiffness coefficient is actively reduced to decrease the load on the execution system by relaxing the tolerance for longitudinal following errors. Simultaneously, the controller constrains the ratio between the virtual damping coefficient and the stiffness coefficient, maintaining a constant damping ratio of the virtual impedance system to ensure the dynamic stability of the control process.

[0011] Finally, the system introduces a cooperative momentum unloading strategy to cope with extreme conditions of attitude actuator saturation. When the attitude risk index exceeds a preset threshold, it indicates that simply adjusting stiffness is no longer sufficient to meet safety requirements, and the cooperative unloading decision module triggers an active defense mechanism. Based on the current saturation state of the control moment gyroscope module, the module calculates the longitudinal buffer distance required to perform the momentum unloading maneuver and superimposes this distance with the vehicle's actual physical length to generate a virtual vehicle length parameter. This parameter generation process employs an asymmetric expansion strategy, adding only the virtual length component to the front or rear of the vehicle, depending on the direction of gyroscope saturation.

[0012] The virtual vehicle length parameter is broadcast to adjacent vehicles in the platoon via a communication link. Upon receiving the parameter, the adjacent vehicle in front accelerates based on collision avoidance logic, while the adjacent vehicle behind decelerates based on following logic, thus creating a longitudinal operational window for the vehicle in physical space. After the window is formed, the vehicle's control execution system performs pulse-type acceleration and deceleration maneuvers, using the inertial torque generated by acceleration and deceleration to reverse-drive the frame angle reset of the control torque gyroscope module, achieving angular momentum unloading. Once the saturation index drops to the safe recovery threshold, the system resets the virtual vehicle length parameter to the actual physical length, resuming normal platooning.

[0013] This invention provides a self-balancing multi-vehicle platooning control system based on attitude safety constraints. It offers the following advantages: 1. This invention resolves the contradiction between following accuracy and attitude stability of self-balancing vehicles under strong coupling characteristics by establishing a negative correlation mapping mechanism between attitude risk index and formation control stiffness parameters. In the risk warning stage where the execution system tends to saturate, the system automatically reduces the virtual spring stiffness and actively relaxes the tolerance for the following error of the preceding vehicle, thereby reducing the dynamic response requirements of the hub motor. It sacrifices instantaneous following accuracy to gain attitude control margin, effectively preventing attitude instability induced by excessive pursuit of close following.

[0014] 2. This invention proposes a cooperative momentum unloading strategy based on virtual vehicle length parameters, which overcomes the limitation that a single vehicle cannot perform large-scale maneuvers in a tight formation to restore the attitude actuator capability. By dynamically adjusting the space occupied by the vehicle at the logical level through an asymmetric expansion strategy, it guides adjacent vehicles to actively accelerate and decelerate, thereby creating a longitudinal operation window. This allows the vehicle to perform pulse-type acceleration and deceleration maneuvers while maintaining formation continuity. It also enables online reset and cyclic use of the angular momentum of the control torque gyroscope module, ensuring the long-term reliability of the system.

[0015] 3. This invention constructs a three-dimensional attitude safety envelope that integrates the vehicle's motion state and the actuator saturation index, achieving precise quantification of vehicle instability risk. Unlike traditional control that relies solely on attitude angle thresholds, this system incorporates the physical limits of the frame angle of the control torque gyroscope module into the evaluation system. Through hierarchical control of the absolute safety zone, risk warning zone, and instability critical zone, it can identify potential risks caused by the attenuation of actuator capabilities in advance, ensuring the feasibility of formation control strategies within physical constraints. Attached Figure Description

[0016] Figure 1 This is a schematic diagram of the self-balancing multi-vehicle platooning control system based on attitude safety constraints of the present invention; Figure 2 This is a diagram of the internal control logic architecture of the vehicle-mounted main control unit of the present invention; Figure 3 This is a flowchart illustrating the attitude safety envelope and risk quantification process of the present invention. Figure 4 This is a schematic diagram of the risk-adaptive virtual variable stiffness impedance control logic of the present invention; Figure 5 This is a timing flowchart of the cooperative momentum unloading strategy based on virtual vehicle length expansion of the present invention; Figure 6 This is the formation coordination control process under typical climbing conditions of the present invention.

[0017] Among them, 100 is the self-balancing vehicle; 200 is the execution system; 201 is the hub motor module; 202 is the control torque gyroscope module; 300 is the sensing system; 301 is the inertial measurement unit; 302 is the wheel speed sensor; 303 is the status encoder; 400 is the communication system; 401 is the vehicle-mounted V2V communication module; 500 is the vehicle-mounted main control unit; 501 is the status monitoring module; 502 is the risk assessment module; 503 is the variable stiffness controller; and 504 is the collaborative unloading decision module. Detailed Implementation

[0018] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] See attached document Figure 1The present invention provides a multi-vehicle platooning control system for self-balancing vehicles based on attitude safety constraints, comprising: multiple self-balancing vehicles 100 with two-wheel self-balancing chassis, an execution system 200, a sensing system 300, a communication system 400 and an on-board main control unit 500 installed on each self-balancing vehicle 100.

[0020] The self-balancing vehicle 100 achieves upright posture and movement via a two-wheeled self-balancing chassis. This chassis comprises two coaxially distributed drive wheels. This structure is an underactuated, nonholonomic constraint system, where the attitude angle and longitudinal acceleration are dynamically coupled. The execution system 200 generates driving torque and attitude control torque, including hub motor modules 201 mounted in each of the two drive wheels and a control torque gyroscope module 202 mounted inside the vehicle body.

[0021] The hub motor module 201 receives longitudinal control commands from the on-board main control unit 500 and outputs drive torque to drive the self-balancing vehicle 100 to generate longitudinal acceleration. The control torque gyroscope module 202 maintains the static and dynamic balance of the self-balancing vehicle 100. It includes a high-speed rotating rotor, a universal frame for supporting the rotor, and a frame motor for driving the frame rotation. The control torque gyroscope module 202 utilizes the principle of conservation of angular momentum to generate precession gyroscopic torque through the rotation of the frame motor, which acts directly on the vehicle body to resist pitch or roll disturbances.

[0022] In a preferred embodiment, the control moment gyroscope module 202 adopts a paired scissor configuration, comprising two gyroscope units rotating in opposite directions and controlled by a frame linkage. This configuration is set to generate the required longitudinal pitch moment while simultaneously canceling out the lateral moment components of the two gyroscope units, thereby eliminating parasitic interference moments on the vehicle's yaw or roll direction during precession. Alternatively, in a single-frame configuration, the onboard main control unit 500 generates differential torque by controlling the hub motor module 201 to actively compensate for the yaw coupling torque caused by the precession of the single-frame control moment gyroscope.

[0023] The sensing system 300 is used to collect vehicle motion state data. It includes an inertial measurement unit 301 located at the center of the vehicle body, wheel speed sensors 302 located at the drive wheel axles, and a state encoder 303 located within the control torque gyroscope module 202. The inertial measurement unit 301 measures the vehicle's three-axis attitude angles in real time, including pitch, roll, and yaw angles, as well as the corresponding three-axis angular velocities. The wheel speed sensors 302 measure the rotational speed of the drive wheels, thereby calculating the vehicle's longitudinal velocity and displacement. The state encoder 303 measures the deflection angle of the frame relative to the vehicle body and the real-time rotational speed of the rotor in the control torque gyroscope module 202.

[0024] The communication system 400 includes an onboard V2V communication module 401, used to establish low-latency wireless data links between the self-balancing vehicles 100 in the platoon. The onboard V2V communication module 401 follows DSRC or C-V2X communication protocols and is used to periodically broadcast its own status information and receive status information from neighboring vehicles. This status information includes at least the vehicle's longitudinal position coordinates, longitudinal velocity, acceleration, and attitude risk index and virtual vehicle length parameters calculated by the onboard main control unit 500.

[0025] The onboard main control unit 500 is connected to the execution system 200, the perception system 300, and the communication system 400 via a bus. The onboard main control unit 500 runs an attitude safety constraint algorithm and formation cooperative control logic. It is configured to calculate the attitude risk index based on the data collected by the perception system 300, adjust the stiffness parameters of the formation control based on the attitude risk index, and generate virtual vehicle length parameters when specific conditions are met and send them through the communication system 400.

[0026] See attached document Figure 2 The vehicle-mounted main control unit 500 includes a status monitoring module 501, a risk assessment module 502, a variable stiffness controller 503, and a collaborative unloading decision module 504.

[0027] The state monitoring module 501 is connected to the sensing system 300 and is used to receive and process raw data from the sensors to calculate the vehicle's real-time motion state vector. This state vector includes the vehicle's pitch angle. Pitch angular velocity and the frame angle of the control torque gyroscope module 202 .

[0028] The risk assessment module 502 is connected to the state monitoring module 501 and is used to establish an attitude safety envelope based on the physical characteristics of the control torque gyroscope module 202, and to quantitatively calculate the attitude risk index. The risk assessment module 502 first calculates the gyroscope saturation index of the control torque. This index represents the current framework angle. With the maximum permissible mechanical deflection angle The degree of similarity.

[0029] Saturation Index The calculation is as follows: ; in: For the current frame angle; The maximum allowable mechanical deflection angle is preset for the control torque gyroscope module 202.

[0030] Subsequently, the risk assessment module 502 is based on pitch angle deviation and saturation index. Calculate the comprehensive attitude risk index This calculation comprehensively considers kinematic deviation and dynamic saturation, when the saturation index... When the attitude risk index exceeds the preset linear region threshold, It exhibits a rapid, non-linear increase to reflect the instability risk faced by the sensing system 300.

[0031] The variable stiffness controller 503 is connected to the risk assessment module 502 and is used to generate longitudinal control commands for the hub motor module 201. The variable stiffness controller 503 adopts an impedance control strategy to model the longitudinal following relationship between the self-balancing vehicle 100 and the preceding vehicle as a virtual spring and damping system with variable parameters.

[0032] The variable stiffness controller 503 adjusts the received attitude risk index. Adjust the stiffness coefficient of the virtual spring in real time Stiffness coefficient Posture Risk Index They show a negative correlation. When the attitude risk index... When in the low range, the stiffness coefficient Maintaining the nominal stiffness value ensures the vehicle closely follows; when the attitude risk index... When the height increases, the stiffness coefficient Automatically reduce the control weight of position error, thereby reducing the acceleration and deceleration actions of the hub motor module 201 caused by correcting position error.

[0033] The collaborative unloading decision module 504 is connected to the risk assessment module 502 and the communication system 400, and is used for attitude risk index... or saturation index When the speed exceeds a preset safety threshold, a cooperative momentum unloading strategy is triggered. The cooperative unloading decision module 504 calculates the required momentum recovery maneuver distance and generates virtual vehicle length parameters accordingly. .

[0034] Virtual vehicle length parameters The calculation is as follows: ; in: The actual physical length of the self-balancing vehicle 100; Based on the current saturation index The calculated longitudinal buffer distance required to perform the momentum unloading maneuver.

[0035] The collaborative unloading decision module 504 will update the virtual vehicle length parameters. The virtual vehicle length parameter is broadcast to adjacent vehicles in the platoon via the communication system 400. When the leading or following vehicle in the platoon receives this virtual vehicle length parameter... After the changes, its internal anti-collision or following algorithm adjusts its longitudinal position based on the new virtual boundary, thereby providing a longitudinal operation window for the vehicle that triggers unloading in physical space. Within the formed window, the collaborative unloading decision module 504 controls the hub motor module 201 to perform pulse-type acceleration and deceleration maneuvers, using the generated inertial torque to drive the frame angle of the control torque gyroscope module 202. Reset.

[0036] See attached document Figure 3 This process aims to construct a dynamic attitude safety boundary by real-time monitoring of the vehicle's motion state and the physical limits of the actuators, and to quantify the degree to which the current state deviates from the core safety area, thereby generating an attitude risk index to guide the switching of control strategies.

[0037] The risk quantification method first performs data acquisition and preprocessing steps. The onboard main control unit periodically reads the three-axis angular velocities and Euler angles of the vehicle body from the inertial measurement unit, and then performs Kalman filtering to eliminate sensor noise. Simultaneously, it reads the current frame angular position from the status encoder of the control torque gyroscope module. and rotor speed The system is set to synchronize the sampling period with the control period to ensure the real-time nature of risk assessment.

[0038] Next, the system calculates the real-time output capability of the control moment gyroscope based on the collected data. The control moment gyroscope generates gyroscopic torque by utilizing the precession of the angular momentum vector of a high-speed rotating rotor. The magnitude of its output torque is related to the vector product of the frame angular velocity and the rotor angular momentum, and is subject to the geometric constraints of the current frame angular position. When the frame angle... When approaching the mechanical limit or singularity point of ±90°, the determinant of the Jacobian matrix of the effective output torque approaches zero, meaning that regardless of how the frame motor accelerates, the effective restoring torque output to the vehicle body will decay to zero. Therefore, the system defines the control torque gyro saturation index at the current moment. This index is the absolute value of the current frame angle and the maximum permissible mechanical deflection angle. The ratio of . The value is the result of subtracting the safety margin from the pre-calibrated mechanical hard limit angle.

[0039] The saturation index was calculated. Subsequently, the system constructs a multi-dimensional attitude safety envelope. This envelope is defined by the vehicle pitch angle. Pitch angular velocity and the control torque gyroscope saturation index In the three-dimensional state space formed.

[0040] The security envelope is divided into three regions: The first region is the absolute safety zone, corresponding to a small range of fluctuations in both pitch angle and angular velocity, and a control moment gyroscope saturation index. When the value is low (e.g., less than 0.5), the vehicle has full attitude recovery capability. The second area is a risk warning zone, corresponding to an increase in pitch angle or angular velocity, or an increase in the control torque gyroscope saturation index. When entering the high range (e.g., 0.5 to 0.85), attitude control is effective, but the actuator margin is reduced. The third region is the instability critical zone, corresponding to the control moment gyroscope saturation index. When the speed approaches 1.0, or the vehicle's motion state is close to the overturning boundary, it is difficult to maintain balance by relying solely on the control moment gyroscope. It is necessary to use the gravitational torque generated by the longitudinal acceleration of the wheels for assistance.

[0041] Based on the above regional division, the system calculates the comprehensive attitude risk index. The index is calculated using a piecewise nonlinear weighted function model. When the vehicle is in the absolutely safe zone, Primarily determined by the vehicle pitch angle deviation, its value is low and changes gradually, reflecting normal road disturbances. When the vehicle enters the risk warning zone, the function model introduces a saturation index. As a gain factor, it amplifies the pitch angle deviation. Follow The increase in saturation index is linear. When the vehicle enters the instability critical region, the function model switches to exponential growth mode. At this point, even if the pitch angle deviation is small, as long as the saturation index increases... Approaching 1.0, It will also quickly spike to a normalized high value (e.g., close to 1.0), thereby sending a high-priority intervention request to subsequent control modules.

[0042] In addition, the risk quantification method also includes the rotor speed. Monitoring. If the rotor speed is detected to be lower than a preset percentage of the rated operating speed, the system will automatically introduce a speed compensation coefficient to adjust the attitude risk index. A correction is made so that, under the same saturation, the lower the rotational speed, the higher the calculated risk index, in order to reflect the weakening control capability caused by the decrease in angular momentum modulus.

[0043] The final generated attitude risk index The data is output in real time to the variable stiffness controller and the collaborative unloading decision module, serving as the sole basis for the system to adjust virtual impedance parameters and trigger formation topology reconfiguration. Through this risk quantification based on the physical limits of the actuators, the system can predict risks before actual attitude instability, solving the lag problem caused by traditional methods that rely solely on tilt angle thresholds.

[0044] See attached document Figure 4 This control mode is mainly applied to the steady-state process of vehicles traveling in formation. It aims to solve the resource competition problem between longitudinal position maintenance and attitude stability in strongly coupled systems. By dynamically adjusting the hardness of the longitudinal following characteristics, it achieves active surrender of attitude safety.

[0045] In traditional vehicle platooning control, a rigid position error feedback mechanism is typically used. This means that once a vehicle spacing error occurs, the controller outputs the maximum possible acceleration or deceleration command to eliminate the error. However, for underdriven two-wheeled self-balancing vehicles, longitudinal acceleration must be accompanied by a change in the vehicle's pitch angle (acceleration leads to forward lean, deceleration leads to backward lean). If rigid position correction commands are forcibly executed when the vehicle's attitude risk is high, it can easily induce attitude divergence. Therefore, this embodiment employs an impedance control architecture, treating the longitudinal interaction between the vehicle and the preceding vehicle (or virtual leader point) as an equivalent virtual mechanical impedance system.

[0046] This virtual impedance system consists of a virtual spring and a virtual damper connected in parallel. The virtual spring generates a restoring force proportional to the position error to maintain the desired vehicle spacing; the virtual damper generates a damping force proportional to the relative speed error to suppress longitudinal oscillations within the formation. In this case, the output of the vehicle's longitudinal motion controller is no longer directly determined by the position error, but rather follows the dynamic response of this virtual impedance model.

[0047] To achieve attitude safety constraints, the system introduces an attitude risk index-based approach. The adaptive parameter adjustment mechanism adjusts the stiffness coefficient of the virtual spring. Real-time dynamic mapping is performed. This mapping mechanism establishes a monotonically decreasing function relationship, which maps the input attitude risk index. Stiffness coefficient converted into output .

[0048] Specifically, the system has two preset stiffness reference values: nominal high stiffness value and nominal high stiffness value. and safety low stiffness value When the attitude risk index When the value is below the preset safety threshold (indicating stable vehicle attitude and sufficient control torque gyroscope momentum), the system output is close to... The stiffness coefficient is [value missing]. At this point, the virtual connection exhibits the characteristics of a stiff spring, the controller is highly sensitive to position errors, and the hub motors actively respond to minute changes in spacing, prioritizing the tightness and accuracy of the formation.

[0049] When the posture risk index The coefficient gradually increases (indicating that the vehicle has encountered a disturbance or the control moment gyroscope is approaching saturation), and the mapping function drives the stiffness coefficient. Towards Smooth decay. At this point, the virtual connection exhibits soft spring characteristics, and the controller's sensitivity to position errors decreases. This means the system allows for larger elastic float errors in the vehicle's longitudinal position, thereby significantly reducing the abrupt acceleration and deceleration actions performed by the hub motors to correct position.

[0050] By employing this stiffness-softening strategy, the system physically suppresses the bandwidth of longitudinal acceleration commands, effectively acting as a low-pass filter to eliminate position correction commands that would trigger high-frequency attitude oscillations. This approach prioritizes the use of hub motor power resources and vehicle pitch freedom for the attitude balance control loop, ensuring that the vehicle can temporarily lag behind to avoid tipping over in high-risk situations.

[0051] Furthermore, to prevent abrupt changes in stiffness coefficients from causing jumps in control commands, a rate-of-change limitation is introduced into the parameter adjustment mechanism. This applies regardless of the attitude risk index. How drastic fluctuations occur, stiffness coefficient The change in velocity at every moment is limited to a preset slope range, ensuring a smooth and continuous transition from hard connections to soft connections. Simultaneously, to guarantee formation safety, a virtual damping coefficient is used. It is usually kept constant or adjusted synchronously with the stiffness coefficient to ensure that the system still has a sufficient damping ratio when the stiffness decreases, thus preventing underdamped oscillations.

[0052] Specifically, in order to maintain the stability of the system's dynamic response, the virtual damping coefficient... With stiffness coefficient Follow the critical damping or preset damping ratio Constraint relationships. When the stiffness coefficient As the risk index decreases, the variable stiffness controller 503 synchronously adjusts the virtual damping coefficient. To satisfy (in For the equivalent quality of the system, The damping ratio is preset (e.g., 0.707). By keeping the damping ratio constant, it is ensured that the longitudinal movement of the vehicle is always in a convergent state with no overshoot or slight overshoot during the transition from a rigid connection to a flexible connection, thus preventing low-frequency longitudinal oscillations induced by reduced stiffness.

[0053] The final control quantity (target acceleration or torque) output by the variable stiffness controller is sent to the underlying motor drive unit to drive the hub motor to perform actions. This control quantity is a function of position error, relative velocity error, and the current real-time stiffness coefficient and damping coefficient. Through the above mechanism, this embodiment realizes the dynamic switching of priority between formation position commands and individual vehicle attitude commands at the algorithm level, without the need for complex logical judgment statements, but rather internalized as the continuous evolution of physical model parameters.

[0054] See attached document Figure 5 This control mode belongs to the system's ultimate safety defense layer and is triggered only when variable stiffness impedance control cannot suppress risk accumulation and the control moment gyroscope is about to enter an irreversible saturation failure state. Its core lies in using the V2V communication network to modify the vehicle's geometric properties in the control domain and actively creating physical operating space by utilizing the collision avoidance logic among formation members, thereby achieving physical unloading of the momentum of a single vehicle.

[0055] When the status monitoring module determines that a certain self-balancing vehicle (let's call it the number 1) is in operation... (Car) Control Torque Gyro Saturation Index If the risk level continuously exceeds the preset high-risk warning threshold (e.g., 0.85) and continues to rise after stiffness adjustment, the system will immediately activate the cooperative momentum unloading procedure.

[0056] Before activating the cooperative momentum unloading procedure, the cooperative unloading decision module 504 must perform an energy feasibility check. The system detects the remaining charge and maximum discharge power capability of the vehicle's battery pack. Virtual vehicle length expansion and momentum reset are only performed if the remaining charge is higher than the safe maneuver threshold and the battery can support the instantaneous peak power required for pulse maneuvering. If the charge check fails, the system will prohibit triggering cooperative unloading and instead execute an emergency degrading strategy, such as controlling the deployment of telescopic auxiliary support wheels, switching the vehicle to a statically stable support mode, and requesting to leave the formation.

[0057] First of all, the The vehicle's onboard main control unit, based on the current control torque gyroscope frame angle deviation and rotor angular momentum, inversely calculates the longitudinal acceleration and deceleration maneuver required to reset the frame angle to zero. This maneuver typically manifests as a specific velocity-time profile (e.g., acceleration followed by deceleration, using inertial torque to drive the frame in the opposite direction). The unit integrates this profile to calculate the required longitudinal displacement, and then adds a preset safety redundancy to determine the final required longitudinal buffer distance. .

[0058] Subsequently, the system generates the first The vehicle's virtual geometric parameters. The onboard main control unit will determine the vehicle's physical length. Compared with the calculated longitudinal buffer distance By superimposing and synthesizing, a new parameter is created—the virtual vehicle length. Furthermore, the length of the virtual vehicle The expansion exhibits directional selectivity (asymmetry). The cooperative unloading decision module 504 determines the motion direction required for momentum reset maneuvers based on the current control torque gyroscope saturation direction. If a forward acceleration maneuver is required to reset the pitch saturation, only the virtual length component of the front of the vehicle is increased. The virtual length of the rear remains unchanged; if a rearward deceleration maneuver is required to reset the forward tilt saturation, only the virtual length component of the rear of the vehicle is increased. .

[0059] This asymmetric expansion strategy ensures that the formation undergoes topological deformation only on the necessary side, minimizing the impact on the overall queue's movement efficiency. This parameter Numerically larger than the vehicle's actual physical size, its physical significance lies in defining the exclusion domain occupied by the vehicle in the formation control algorithm. The vehicle communicates with the preceding vehicle in the queue via a V2V communication module using high-priority messages. (car) and the car behind ( The vehicle broadcasts the updated virtual vehicle length parameters and simultaneously sends the momentum unloading request flag.

[0060] Received the After the vehicle broadcasts the virtual vehicle length expansion information, the cooperative adaptive cruise control controllers of adjacent vehicles automatically respond to this topology change. For the vehicle in front ( (Vehicle), whose rear radar or virtual perception module determines the rearmost vehicle. The car's virtual boundary suddenly moved forward, encroaching on the preset safe following distance. Based on collision avoidance and distance maintenance logic, the vehicle in front automatically executed an acceleration command to increase the distance from the virtual boundary. For the vehicle behind ( (Car), its forward perception logic determines the first one in front. The vehicle's virtual boundary suddenly shifted backward, also intruding on the safe following distance. Based on the cooperative adaptive cruise control following logic, the following vehicle automatically executed a deceleration command to make room. Through the coordinated actions of the leading vehicle accelerating forward and the following vehicle decelerating backward, the convoy reached its safe following distance in the [missing information]. The car is physically stretched in the longitudinal direction, forming a length equal to... The vertical physical window area.

[0061] Once the sensing system confirms that a physical window area has been formed and meets safety requirements, the first... The vehicle temporarily releases the formation position lock constraint and enters a momentum reset maneuver. In this state, the hub motors no longer respond to formation following commands but instead execute pre-calculated pulse-like acceleration and deceleration maneuvers. For example, if a rearward pitch torque is needed to correct forward pitch saturation, the vehicle will perform a sharp acceleration maneuver within the clearance zone. Utilizing the d'Alembert inertial force generated during acceleration, a counter-torque is applied to the control moment gyro frame, forcing the frame angle, which has already deflected to its limit position, to overcome the gyroscopic effect and return to near zero. The entire maneuver is strictly confined to the physical clearance zone left by the vehicles in front and behind, ensuring no physical collision with adjacent vehicles.

[0062] When the control torque gyroscope frame angle is successfully reset, the saturation index... After falling below the safe recovery threshold, the first The car will display the virtual vehicle length. Reset to the actual physical length The momentum unloading request is then cancelled. Adjacent vehicles perceive the contraction of the virtual boundary, the vehicle in front decelerates and the vehicle behind accelerates, eliminating the accumulated spacing error and allowing the formation to converge smoothly like an accordion playing back, restoring the initial close cruising state.

[0063] This strategy transforms the internal actuator maintenance requirements (momentum unloading) of a single vehicle into external geometric constraints (virtual vehicle length) visible to the entire fleet through virtual-real mapping. It utilizes existing cooperative adaptive cruise collision avoidance logic to achieve complex cooperative control and solves the physical problem that underactuated vehicles cannot recover momentum on their own in confined spaces.

[0064] See attached document Figure 6 This embodiment describes a specific dynamic scenario and elaborates in detail how the system sequentially activates variable stiffness impedance control and virtual vehicle length expansion strategy when dealing with continuous external disturbances in order to maintain the overall safety of the formation.

[0065] Suppose a convoy of three self-balancing vehicles (labeled V1, V2, and V3) is cruising at a constant speed on a straight road. At time T0, the convoy enters a long, continuous uphill section. Influenced by the downward component of gravity along the slope, the vehicles are continuously subjected to a pitching torque disturbance. To maintain the vehicles' dynamic balance, the actuators must continuously output a counter-torque. Since the hub motors need to maintain a constant cruising speed, some low-frequency attitude balancing loads are distributed to the control moment gyroscope, causing it to continuously output precessing torque to counteract the pitching tendency. At this time, the frame angle of the control moment gyroscope gradually deflects to one side from zero, and the saturation index slowly increases accordingly.

[0066] During the time interval from T0 to T1, as the slope continues, the deflection of the control torque gyro frame of vehicle V2 gradually increases, causing the calculated attitude risk index to enter the risk warning zone. At this time, the onboard main control unit of vehicle V2 activates the virtual variable stiffness impedance control mode. According to the preset inverse correlation mapping logic, the system automatically reduces the virtual stiffness coefficient in the longitudinal following control loop. In macroscopic terms, vehicle V2 no longer forcibly maintains a precisely set distance from the preceding vehicle V1, but allows itself to have a certain longitudinal positional lag relative to the preceding vehicle V1 (i.e., the distance between the vehicles is passively increased). This soft connection state reduces the need for the hub motors to frequently adjust torque to catch up with the position, thereby avoiding further deterioration of the vehicle's backward pitching tendency due to the motor's violent acceleration, and prioritizing the allocation of limited control authority to the attitude balance system.

[0067] At time T1, assuming the gradient increases further or a headwind is encountered, the control moment gyroscope frame angle of vehicle V2 continues to accumulate and approaches the mechanical limit (e.g., reaching 85 degrees), and the saturation index exceeds the high-risk warning threshold. At this point, simple variable stiffness adjustment can no longer curb the saturation trend. If no measures are taken, the control moment gyroscope will enter a singular state, causing vehicle instability. The system then triggers a cooperative momentum unloading strategy. The main control unit of vehicle V2 first calculates the reverse inertial torque required to pull the control moment gyroscope frame angle back from 85 degrees to 0 degrees, and based on this, calculates the longitudinal buffer distance (e.g., 1.5 meters) required to perform a rapid acceleration and deceleration pulse maneuver.

[0068] Next, vehicle V2 superimposes this buffer distance onto its own physical length to generate an expanded virtual vehicle length, which is then instantly broadcast to the preceding vehicle V1 and the following vehicle V3 via the V2V network. In the logic of the preceding vehicle V1, the perceived virtual boundary of vehicle V2 suddenly extends forward; to avoid a rear-end collision, the preceding vehicle V1 automatically accelerates to create distance. In the logic of the following vehicle V3, the perceived virtual boundary of vehicle V2 suddenly extends backward; to maintain a safe following distance, the following vehicle V3 automatically decelerates and moves backward. This coordinated process rapidly creates a longitudinal operating window zone of approximately 1.5 meters in length for vehicle V2 in physical space.

[0069] At time T2, after confirming the formation of the clearance zone, vehicle V2 performs a predetermined momentum reset maneuver within the clearance zone. The hub motors execute a short, sharp deceleration (or reverse braking) command, causing the vehicle body to generate a strong forward tilting torque due to inertia. This physical torque forcibly overcomes the gyroscopic effect, driving the frame axis of the control torque gyroscope to rotate in the opposite direction, rapidly returning it from a near-saturation 85-degree position to a midpoint near 0 degrees. Although this process causes a sharp fluctuation in the vehicle's position, a collision will not occur because it is within the pre-established clearance zone and the vehicles in front and behind are already in a avoidance maneuver.

[0070] At time T3, it was detected that the control moment gyroscope frame angle had been successfully reset and the saturation index had dropped to a safe range. Vehicle V2 restored the broadcast virtual vehicle length to the actual physical length and gradually restored the stiffness coefficient to the nominal value. Vehicles V1 and V3, sensing the virtual boundary contraction, performed deceleration and acceleration respectively, driven by the cooperative adaptive cruise algorithm, gradually eliminating the extra spacing generated at time T1. Finally, at time T4, the three vehicles returned to a tight, steady-state cruise formation, completing online momentum maintenance for each vehicle without stopping or disbanding the formation. This complete process embodies a hierarchical defense mechanism from micro-parameter adjustment to macro-topology reconstruction.

Claims

1. A self-balancing multi-vehicle platooning control system based on attitude safety constraints, characterized in that, Including multiple self-balancing vehicles; Each of the self-balancing vehicles is equipped with an execution system, a sensing system, a communication system, and an on-board main control unit; The execution system is used to generate driving torque and attitude control torque; the sensing system is used to collect vehicle motion state data; the communication system is used to establish data links between the self-balancing vehicles in the formation; the on-board main control unit is connected to the execution system, the sensing system and the communication system respectively. The vehicle-mounted main control unit is configured to perform the following operations: The attitude risk index is calculated based on the data collected by the perception system; the stiffness parameters of the formation control are adjusted based on the attitude risk index; when preset conditions are met, virtual vehicle length parameters are generated and sent to adjacent vehicles through the communication system. The virtual vehicle length parameters are used to trigger the cooperative momentum unloading strategy and control the execution system to perform momentum unloading maneuvers.

2. The self-balancing multi-vehicle platooning control system based on attitude safety constraints according to claim 1, characterized in that, The execution system includes a hub motor module and a control torque gyroscope module; The hub motor module is used to receive longitudinal control commands from the on-board main control unit and output driving torque to drive the self-balancing vehicle. The control torque gyroscope module includes a high-speed rotating rotor, a universal frame for supporting the rotor, and a frame motor for driving the frame to rotate. The control torque gyroscope module is configured to generate precession gyroscope torque by utilizing the principle of conservation of angular momentum through the rotation of the frame motor.

3. The self-balancing multi-vehicle platooning control system based on attitude safety constraints according to claim 1, characterized in that, The sensing system includes an inertial measurement unit, a wheel speed sensor, and a state encoder; The inertial measurement unit is located at the center of the vehicle body and is used to measure the vehicle body's three-axis attitude angles and three-axis angular velocities; The wheel speed sensor is located at the drive wheel axle and is used to measure the rotational speed of the drive wheel; The status encoder is installed in the control torque gyroscope module of the execution system and is used to measure the deflection angle of the frame relative to the vehicle body and the real-time rotational speed of the rotor in the control torque gyroscope module.

4. The self-balancing multi-vehicle platooning control system based on attitude safety constraints according to claim 1, characterized in that, The on-board main control unit includes a status monitoring module, a risk assessment module, a variable stiffness controller, and a collaborative unloading decision module. The state monitoring module is used to calculate the real-time motion state vector of the vehicle; The risk assessment module is used to establish an attitude safety envelope based on the physical characteristics of the execution system and to calculate the attitude risk index. The variable stiffness controller is used to generate longitudinal control commands for the execution system based on the attitude risk index. The collaborative unloading decision module is used to trigger the collaborative momentum unloading strategy when the attitude risk index exceeds a preset threshold, and to calculate the virtual vehicle length parameter.

5. The self-balancing multi-vehicle platooning control system based on attitude safety constraints according to claim 4, characterized in that, The specific method by which the risk assessment module calculates the attitude risk index is as follows: Calculate the saturation index of the control torque gyroscope module in the execution system, whereby the saturation index represents the ratio of the current frame angle to the maximum allowable mechanical deflection angle; Construct a three-dimensional attitude safety envelope that includes vehicle pitch angle, pitch angular velocity, and the saturation index; The three-dimensional attitude safety envelope is divided into an absolute safety zone, a risk warning zone, and an instability critical zone. When the vehicle is in the risk warning zone, the attitude risk index increases linearly with the increase of the saturation index. When the vehicle enters the instability critical zone, the attitude risk index increases exponentially.

6. The self-balancing multi-vehicle platooning control system based on attitude safety constraints according to claim 4, characterized in that, The variable stiffness controller adopts an impedance control strategy to model the longitudinal following relationship between the self-balancing vehicle and the preceding vehicle as a virtual impedance system containing virtual springs and virtual dampers. The variable stiffness controller adjusts the stiffness coefficient of the virtual spring in real time based on the received attitude risk index. The stiffness coefficient is negatively correlated with the attitude risk index; When the attitude risk index is in a low range, the stiffness coefficient remains at the nominal stiffness value. When the attitude risk index increases, the stiffness coefficient decreases.

7. The self-balancing multi-vehicle platooning control system based on attitude safety constraints according to claim 6, characterized in that, The variable stiffness controller is configured to simultaneously adjust the virtual damping coefficient of the virtual damper while adjusting the stiffness coefficient. The variable stiffness controller keeps the damping ratio of the virtual impedance system constant by constraining the relationship between the virtual damping coefficient and the stiffness coefficient.

8. The self-balancing multi-vehicle platooning control system based on attitude safety constraints according to claim 4, characterized in that, The specific method by which the collaborative unloading decision module generates the virtual vehicle length parameters is as follows: The longitudinal buffer distance required to perform the momentum unloading maneuver is calculated based on the saturation state of the control torque gyroscope module in the execution system. The virtual vehicle length parameter is obtained by superimposing the actual physical length of the self-balancing vehicle with the longitudinal buffer distance. The virtual vehicle length parameter is generated using an asymmetric expansion strategy, which increases the virtual length component only at the front or rear of the vehicle according to the saturation direction of the control torque gyroscope module.

9. The self-balancing multi-vehicle platooning control system based on attitude safety constraints according to claim 8, characterized in that, The collaborative unloading decision module broadcasts the virtual vehicle length parameters to adjacent vehicles in the formation via the communication system. After the adjacent vehicles receive the virtual vehicle length parameters, the adjacent vehicle in front executes an acceleration command based on collision avoidance logic, and the adjacent vehicle behind executes a deceleration command based on following logic. The acceleration of the adjacent vehicle in front and the deceleration of the adjacent vehicle behind create a longitudinal operating window in physical space.

10. The self-balancing multi-vehicle platooning control system based on attitude safety constraints according to claim 9, characterized in that, After the longitudinal operation window is formed, the collaborative unloading decision module controls the execution system to perform pulse-type acceleration and deceleration maneuvers; The pulse-type acceleration and deceleration maneuver utilizes the generated inertial torque to drive the frame angle of the control torque gyroscope module to reset; After the saturation index of the control torque gyroscope module drops to the safety recovery threshold, the collaborative unloading decision module resets the virtual vehicle length parameter to the actual physical length.

Images