Amphibious vehicle amphibious working condition autonomous conversion decision-making system and method
By designing an autonomous conversion decision system for amphibious vehicles, using Bayesian network and real-time control, the problems of low operating conditions conversion efficiency and safety hazards in the existing technology are solved, and the autonomous vehicle is achieved in complex environments and safe working conditions change.
Patent Information
- Application Number
- CN202410898798.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-05
- Publication Date
- 2024-11-22
- Estimated Expiration
- 2044-07-05
AI Technical Summary
Existing amphibious vehicles rely on human operation during operating conditions, which are inefficient and have great safety risks, and traditional methods cannot effectively deal with autonomous switching decisions in complex environments.
A decision-making system for autonomous conversion of water and land conditions was designed, combining the vehicle environmental parameter perception module, vehicle operating condition status feedback system and water and land conditions conversion decision-making module, using Bayesian network and real-time control, it integrates the environment parameters and vehicle status based on environmental parameters and vehicle status. Decision-making to achieve safe and efficient working conditions.
It improves the adaptability and safety of autonomous switching of water and land conditions, reduces driver manual operation, and realizes the vehicle's independent decision-making ability in complex environments.
Smart Images

Figure CN118991777A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of intelligent control and driving assistance systems for autonomous driving vehicles, and in particular to a system and method for automatic conversion between water and land working conditions. Background Art
[0002] Amphibious vehicles can travel on land and water and have a wide range of application prospects. However, current amphibious vehicles often switch between operating modes manually, which leads to low efficiency or distraction of the driver, which can easily cause safety hazards. Therefore, a method is needed that can intelligently decide on the operating mode conversion based on the current state of the vehicle and environmental conditions.
[0003] During the vehicle's land-water switching process, it is necessary to perform rule-based switching according to the current land-water state. Considering the complex environment when the vehicle switches from the intermediate state to the land state during the login process, it is impossible to switch completely based on rules, and decision-making in complex environments is required.
[0004] In the face of complex decision-making problems, using fuzzy knowledge for uncertain reasoning can solve problems such as decision-making reasoning under current uncertain conditions. Bayesian network is a method of expressing and reasoning uncertain knowledge that combines probability theory and graph theory. It is currently one of the most effective models in the field of uncertain knowledge representation and reasoning. The Bayesian network model can represent human knowledge through a directed graph and integrate statistical data into the model in the form of conditional probability. In the tidal flat environment where amphibious vehicles are traveling, due to the complex water and land environment, the autonomous switching and safety issues of amphibious vehicles during driving, the Bayesian network has great advantages in dealing with uncertain problems. It can effectively integrate expert prior knowledge and experimental sample data to complete the reasoning of uncertain problems, which can help solve the autonomous decision-making problems of amphibious vehicles.
[0005] Traditional amphibious vehicle switching is mostly based on human experience, and the decision to switch operating conditions is made by visually judging the current distance, whether the chassis is bottomed out, and other conditions. There is no reliable method to fully transform human experience into an autonomous driving solution. At present, the autonomous switching technology of amphibious vehicles is mainly based on simple rules, but these rules cannot cover the switching function of operating conditions in complex situations. Summary of the invention
[0006] In view of this, the present invention provides an amphibious vehicle working condition conversion decision system and method, which divides the water-land switching process into several processes, and performs autonomous water-land working condition switching based on certain rules according to the characteristics of each process, thereby improving the adaptability and safety of autonomous water-land working condition switching. The method can make a comprehensive decision based on information such as the distance from the shore and the water depth to achieve safe and efficient working condition conversion.
[0007] The specific technical solutions are as follows:
[0008] An autonomous water-land working condition conversion decision system for an amphibious vehicle, comprising a vehicle-mounted environmental parameter perception module, a vehicle working condition state feedback system, a water-land working condition conversion decision module, and a water-land working condition conversion execution unit;
[0009] The vehicle-mounted environmental parameter perception module includes a depth meter, a tire force sensor, an inertial navigation system, a front-view camera, and a rear-view camera, which are used to detect vehicle information;
[0010] The vehicle operating state feedback system is used to provide real-time feedback on the current vehicle's water and land operating state, including water operating state, intermediate operating state, and land operating state;
[0011] The water-land working condition conversion decision module decides and outputs the water-land working condition instruction according to the perception result of the environmental parameter perception module and the working condition given by the vehicle working condition state feedback system;
[0012] The water-land working condition conversion execution unit converts the water-land working condition instruction output by the water-land working condition conversion decision module into the action of the vehicle's water actuator or land actuator, thereby realizing the vehicle's water-land working condition conversion function.
[0013] Furthermore, the vehicle-mounted environmental parameter perception module obtains the current water depth by processing the data of the depth meter, obtains the vertical load of the current wheel by processing the data of the tire force sensor, obtains the current distance between the vehicle and the shore by processing the forward-looking camera and the inertial navigation global positioning information, and obtains the current water and land status of the rear of the vehicle by processing the data of the rear-view camera. The water and land status of the rear of the vehicle includes two states: "on water" and "on land".
[0014] A method for autonomously switching between water and land working conditions of an amphibious vehicle, comprising:
[0015] The vehicle's water-land switching process is divided into three stages: on water, in the middle, and on land. Based on the characteristics of each stage and the current water-land working state of the vehicle, the Bayesian network, conversion rules, and real-time control are combined to output the working state conversion instructions.
[0016] Among them, if the vehicle is currently in the water condition, a Bayesian network is constructed to calculate whether to switch to the intermediate condition;
[0017] If the vehicle is currently in an intermediate state, the intermediate state conversion rule is used to determine whether the intermediate state is converted to a water state, a land state, or not converted;
[0018] If the vehicle is currently in the onshore condition, the distance between the vehicle and the shore is used to determine whether to switch to the intermediate condition;
[0019] Furthermore, if the vehicle is currently in a water condition, the decision-making process includes:
[0020] Step S221, constructing a Bayesian network, defining input node variables in the network, including water depth D, vertical load L, and distance R between the vehicle and the shore, and defining all output node variables in the network, which is the current water-land switching instruction S;
[0021] Step S222: Determine the attributes of each node according to the historical statistical results, and determine the probability of the above node attributes according to the historical statistical results, and each probability is represented by the letter P;
[0022] Among them, the distance R between the vehicle and the shore is divided into two attributes {large, small}, and the node attribute probabilities are expressed as P(R = "large") and P(R = "small");
[0023] P(R = "large") + P(R = "small") = 1
[0024] The water depth is divided into two attributes {deep, shallow}, and the node attribute probabilities are expressed as P(D = "shallow") and P(D = "deep");
[0025] P(D=“shallow”)+P(D=“deep”)=1
[0026] The vertical load of the wheel is divided into two attributes {large, small}, and the node attribute probabilities are expressed as P(L = "large") and P(L = "small");
[0027] P(L = "large") + P(L = "small") = 1
[0028] Under the current working condition, the current water-land switching instruction is divided into two attributes: {do not switch, switch to the middle}, and the node attribute probabilities are expressed as P(S = "do not switch") and P(S = "switch to the middle");
[0029] P(S=“do not switch”)+P(S=“switch to the middle”)=1
[0030] Step S223, respectively calculate the probability of different working condition conversion decisions under the condition of given observation data, that is:
[0031]
[0032] Where P(S=s) represents the prior probability that S takes the value of s{“switch to the middle”, “do not switch”}, which is obtained through historical data statistics;
[0033] P(A|S=s) represents the probability of observing A under the condition that S takes the value s. A represents the result of the permutation and combination of the input node variables in step S221 under various attributes. Considering that each input node in step S221 depends independently on S, we have:
[0034] P(A|S=s)=P(D=d|S=s)·P(L=l|S=s)·P(R=r|S=s)
[0035] P(A) is the probability of observing A;
[0036] Among them, d represents the value of D, l represents the value of L, and r represents the value of R;
[0037] Step S224: Select the result with the highest posterior probability as the final decision, that is:
[0038]
[0039] Furthermore, the intermediate state transition rule is determined by the current water and land state of the rear of the vehicle:
[0040] The current land and water state of the rear of the vehicle is "on land". After it changes to "on water", the land and water working state conversion decision module sends a "switch to water" command to the land and water working state conversion execution unit;
[0041] The current land and water state of the rear of the vehicle is "on water". After it becomes "on land", the land and water working condition conversion decision module sends a "switch to land" command to the land and water working condition conversion execution unit.
[0042] Furthermore, if the vehicle is currently on land, when the distance between the vehicle and the shore is less than the threshold D T When , the decision output switches the middle instruction.
[0043] Furthermore, the mapping of the working condition conversion instruction to the execution mechanism specifically includes:
[0044] When the "onshore working condition" is switched to the "intermediate working condition", the water jet propulsion system is turned on, the wave-breaking board is deployed, and all the air intake devices are closed, and the transmission system switches from only land transmission to simultaneous water and land transmission;
[0045] When the "intermediate working condition" is switched to the "on-water working condition", the sonar is turned on, the suspension is shortened, and the transmission system switches from working on both land and water to working on water only;
[0046] When the "aquatic working condition" is switched to the "intermediate working condition", the sonar is turned off, the suspension is lowered, and the transmission system switches from working only on water to working on both land and water;
[0047] When the "intermediate working condition" is switched to the "land working condition", the water jet propulsion is turned off, the wave-breaking board is retracted, the air intake device is opened, and the transmission system switches from simultaneous water and land transmission to land transmission only.
[0048] The beneficial effects of the present invention are as follows:
[0049] 1. The present invention proposes an amphibious vehicle autonomous switching method based on a decision network model to realize the autonomous working state switching function of amphibious vehicles on land and water, thereby improving the adaptability and ease of operation of the vehicle.
[0050] 2. Compared with traditional amphibious vehicles, the present invention can reduce the switching mechanism manually operated by the driver.
[0051] 3. Existing amphibious vehicles mainly rely on manual decision-making to switch between water and land working conditions. The existing method of switching between water and land working conditions through multi-input Bayesian networks is difficult to ensure real-time solution in practical applications due to the large amount of calculation and the large amount of statistical data. In view of the above problems, the present invention combines a rule-based method and a simple Bayesian network-based method to improve decision-making efficiency, thereby realizing autonomous working condition switching of amphibious vehicles.
[0052] 4. The innovation of the present invention lies in the combination of switching rules, Bayesian networks and real-time control, which effectively improves the autonomous ability of amphibious vehicles during the working condition conversion process and provides an innovative solution for the vehicle's autonomous conversion decision-making between water and land working conditions. BRIEF DESCRIPTION OF THE DRAWINGS
[0053] Figure 1 , Block diagram of the autonomous working state switching system of amphibious vehicles based on the decision network model;
[0054] Figure 2 , system operation flow chart;
[0055] Figure 3 , schematic diagram of Bayesian network nodes. DETAILED DESCRIPTION
[0056] The following is combined with Figure 1-3 Expand to describe the technical solution of the present invention, Figure 1 The paper shows the composition of an amphibious vehicle autonomous working state switching system. Figure 2 The autonomous working state switching process of an amphibious vehicle is demonstrated. Figure 3 The Bayesian network node distribution is shown. :
[0057] An amphibious vehicle operating condition conversion decision system includes an environmental parameter perception module, a vehicle operating condition feedback system, an amphibious operating condition conversion decision module, and an amphibious operating condition conversion execution unit;
[0058] The environmental parameter perception module includes a depth meter, a tire force sensor, an inertial navigation system, a front-view camera, and a rear-view camera;
[0059] Furthermore, the environmental parameter perception module obtains the current water depth by processing the data of the depth meter, obtains the current vertical load of the wheel by processing the data of the tire force sensor, obtains the current distance between the vehicle and the shore by processing the forward-looking camera and the inertial navigation global positioning information, and obtains the current water and land status of the rear of the vehicle by processing the data of the rear-looking camera;
[0060] The vehicle operating state feedback system can provide real-time feedback on the current vehicle's land and water operating state, including "land operating state", "water operating state" and "intermediate operating state";
[0061] The water-land working condition conversion decision module takes the distance between the vehicle and the shore, the water depth, the vertical load of the wheels, the current state of the vehicle entering or leaving the water, and the current water-land working condition state as input, and outputs the water-land working condition instruction to the water-land working condition conversion execution unit through the water-land working condition conversion decision module;
[0062] Furthermore, the water and land working state instructions include a "switch to water" instruction, a "switch to land" instruction, a "switch to intermediate state" instruction and a "do not switch" instruction;
[0063] The water-land working condition execution unit takes the water-land working condition conversion instruction as input, and converts it into the action of the vehicle's water actuator or land actuator after processing, so as to realize the vehicle's water-land working condition conversion function. The specific actions include suspension lifting, sonar retraction and extension, water-land transmission system switch, air intake device switch, wave-breaking board retraction and extension, etc.
[0064] The water-land working condition conversion decision method, the specific implementation process includes the following steps:
[0065] Step S1, the vehicle-mounted environmental parameter perception module senses the external environment to obtain the current water depth, vertical load, distance between the vehicle and the shore, and the water and land status of the rear of the vehicle;
[0066] Step S2, the water-land working condition conversion decision module outputs a working condition conversion instruction based on the perception result of the vehicle-mounted environmental parameter perception module;
[0067] Step S3: the water-land working condition conversion execution unit maps the working condition conversion instruction to the execution mechanism to realize the water-land working condition conversion.
[0068] Furthermore, the step S2 includes the following sub-steps:
[0069] Step S21, reading the water and land working state of the vehicle, if the vehicle is currently in the water working state, executing step S22;
[0070] If the vehicle is currently in the intermediate operating state, execute step S23;
[0071] If the vehicle is currently in a land condition, execute step S24;
[0072] Step S22, calculating the conditional probability of the current working condition conversion instruction according to the statistical probability of each input node, and deciding to output the working condition conversion instruction;
[0073] Step S23: outputting a working condition conversion instruction through an intermediate state conversion rule decision according to the information of the environmental parameter sensing module;
[0074] Specifically, the intermediate state conversion rule is judged by the current land and water state of the rear of the vehicle: if the current land and water state of the rear of the vehicle is "on land", after it becomes "on water", the land and water working state conversion decision module sends a "switch to water" instruction to the land and water working state conversion execution unit; if the current land and water state of the rear of the vehicle is "on water", after it becomes "on land", the land and water working state conversion decision module sends a "switch to land" instruction to the land and water working state conversion execution unit;
[0075] Step S24: Determine the distance between the current vehicle and the shore. When the distance between the vehicle and the shore is less than D T Time (D T is the reference distance calculated through multiple tests), the water-land decision conversion decision module sends a "switch middle" instruction to the water-land working condition conversion execution unit;
[0076] Furthermore, the conditional probability calculation method in step S22 includes the following steps:
[0077] Step S221: construct a Bayesian network, define the input node variables in the network, including water depth (D), vertical load (L), and the distance between the vehicle and the shore (R), and define all output node variables in the network as the current water-land switching instruction (S), such as Figure 3 shown.
[0078] Step S222: Determine the attributes of each node according to the historical statistical results, and determine the probability of the above node attributes according to the statistical results, and each probability is represented by the letter P;
[0079] Specifically, the distance between the vehicle and the shore is divided into “large” and “small”, and the node attribute probabilities are expressed as P(R = “large”) and P(R = “small”);
[0080] P(R = "large") + P(R = "small") = 1
[0081] The water depth is divided into "deep" and "shallow", and the node attribute probabilities are expressed as P(D = "shallow") and P(D = "deep");
[0082] P(D=“shallow”)+P(D=“deep”)=1
[0083] The vertical load of the wheel is divided into "large" and "small", and the node attribute probability is expressed as P(L = "large") and P(L = "small");
[0084] P(L = "large") + P(L = "small") = 1
[0085] Under the current working condition, the current water-land switching instruction is divided into "no switching" and "switch to the middle", and the node attribute probabilities are expressed as P(S = "no switching") and P(S = "switch to the middle");
[0086] P(S=“do not switch”)+P(S=“switch to the middle”)=1
[0087] Step S223, respectively calculate the probability of different working condition conversion decisions under the condition of given observation data, that is:
[0088]
[0089] Where P(S=s) represents the prior probability that S takes the value of s{“switch to the middle”, “do not switch”}, which can be obtained through historical data statistics;
[0090] P(A|S=s) represents the probability of observing A under the condition that S takes the value s. A represents the result of the permutation and combination of the input node variables in step S221 under various attributes. Considering that each input node in step S221 depends independently on S, we have:
[0091] P(A|S=s)=P(D=d|S=s)·P(L=l|S=s)·P(R=r|S=s)
[0092] P(A) is the probability of observing A.
[0093] Step S224: Select the result with the highest posterior probability as the final decision, that is:
[0094]
[0095] The step S3 comprises the following sub-steps:
[0096] Step S31, the water-land working condition conversion execution unit determines the working condition conversion direction according to the current working condition. If the "land working condition" is switched to the "intermediate working condition", step S32 is executed; if the "intermediate working condition" is switched to the "water working condition", step S33 is executed; if the "water working condition" is switched to the "intermediate working condition", step S34 is executed; if the "intermediate working condition" is switched to the "land working condition", step S35 is executed;
[0097] Step S32, turning on the water jet propulsion device, unfolding the wave-breaking board, and closing all the air intake devices, and the transmission system switches from only land transmission to water and land transmission at the same time;
[0098] Step S33, turning on the sonar, shortening the suspension, and switching the transmission system from operating on both land and water to operating only on water;
[0099] Step S34, turn off the sonar, lower the suspension, and switch the transmission system from working only on water to working on both land and water;
[0100] Step S35, turn off the water jet propulsion device, retract the wave-breaking board, open the air intake device, and switch the transmission system from simultaneous water and land transmission to only land transmission operation.
[0101] This embodiment is only a specific implementation of the present invention and does not limit the present invention.
Claims
1. An autonomous water-land working state switching decision system for an amphibious vehicle, characterized by: It includes a vehicle-mounted environmental parameter perception module, a vehicle operating status feedback system, a water-land operating condition conversion decision module, and a water-land operating condition conversion execution unit; The vehicle-mounted environmental parameter perception module includes a depth meter, a tire force sensor, an inertial navigation system, a front-view camera, and a rear-view camera, which are used to detect vehicle information; The vehicle operating state feedback system is used to provide real-time feedback on the current vehicle's water and land operating state, including water operating state, intermediate operating state, and land operating state; The water-land working condition conversion decision module decides and outputs the water-land working condition instruction according to the perception result of the environmental parameter perception module and the working condition given by the vehicle working condition state feedback system; The water-land working condition conversion execution unit converts the water-land working condition instruction output by the water-land working condition conversion decision module into the action of the vehicle's water actuator or land actuator, thereby realizing the vehicle's water-land working condition conversion function.
2. The autonomous water-land working condition conversion decision system for an amphibious vehicle according to claim 1 is characterized by: The vehicle-mounted environmental parameter perception module obtains the current water depth by processing the data of the depth meter, obtains the vertical load of the current wheel by processing the data of the tire force sensor, obtains the current distance between the vehicle and the shore by processing the forward-looking camera and the inertial navigation global positioning information, and obtains the current water and land status of the rear of the vehicle by processing the data of the rear-looking camera, and the water and land status of the rear of the vehicle includes the water state and the land state.
3. A method for autonomously switching between water and land working conditions of an amphibious vehicle, characterized in that: The vehicle's water-land switching process is divided into three stages: on water, in the middle, and on land. Based on the characteristics of each stage and the current water-land working state of the vehicle, the Bayesian network, conversion rules, and real-time control are combined to output the working state conversion instructions. Among them, if the vehicle is currently in the water condition, a Bayesian network is constructed to calculate whether to switch to the intermediate condition; If the vehicle is currently in an intermediate state, the intermediate state conversion rule is used to determine whether the intermediate state is converted to a water state, a land state, or not converted; If the vehicle is currently in the onshore condition, the distance between the vehicle and the shore is used to determine whether to switch to the intermediate condition.
4. The method for autonomously switching between water and land working conditions of an amphibious vehicle according to claim 3 is characterized by: If the vehicle is currently in water conditions, the decision-making process includes: Step S221, constructing a Bayesian network, defining input node variables in the network, including water depth D, vertical load L, and distance R between the vehicle and the shore, and defining all output node variables in the network, which is the current water-land switching instruction S; Step S222: Determine the attributes of each node according to the historical statistical results, and determine the probability of the above node attributes according to the historical statistical results, and each probability is represented by the letter P; Among them, the distance R between the vehicle and the shore is divided into two attributes {large, small}, and the node attribute probabilities are expressed as P(R = "large") and P(R = "small"); P(R = "large") + P(R = "small") = 1 The water depth is divided into two attributes {deep, shallow}, and the node attribute probabilities are expressed as P(D = "shallow") and P(D = "deep"); P(D="shallow")+P(D="deep")=1 The vertical load of the wheel is divided into two attributes {large, small}, and the node attribute probabilities are expressed as P(L = "large") and P(L = "small"); P(L = "large") + P(L = "small") = 1 Under the current working condition, the current water-land switching instruction is divided into two attributes: {do not switch, switch to the middle}, and the node attribute probabilities are expressed as P(S = "do not switch") and P(S = "switch to the middle"); P(S = "do not switch") + P(S = "switch to the middle") = 1 Step S223, respectively calculate the probability of different working condition conversion decisions under the condition of given observation data, that is: Where P(S=s) represents the prior probability that S takes the value of s{"switch to the middle", "do not switch"}, which is obtained through historical data statistics; P(A|S=s) represents the probability of observing A under the condition that S takes the value s. A represents the result of the permutation and combination of the input node variables in step S221 under various attributes. Considering that each input node in step S221 depends independently on S, we have: P(A|S=s)=P(D=d|S=s)·P(L=l|S=s)·P(R=r|S=s) P(A) is the probability of observing A; Among them, d represents the value of D, l represents the value of L, and r represents the value of R; Step S224: Select the result with the highest posterior probability as the final decision, that is:
5. The method for autonomously switching between water and land working conditions of an amphibious vehicle according to claim 3 or 4, characterized in that: The intermediate state transition rules are as follows: The current land and water state of the rear of the vehicle is "on land". After it changes to "on water", the land and water working state conversion decision module sends a "switch to water" instruction to the land and water working state conversion execution unit; The current land and water state of the rear of the vehicle is "on water". After it becomes "on land", the land and water working condition conversion decision module sends a "switch to land" command to the land and water working condition conversion execution unit.
6. The method for autonomously switching between water and land working conditions of an amphibious vehicle according to claim 5 is characterized by: If the vehicle is currently on land, when the distance between the vehicle and the shore is less than the threshold D T When , the decision output switches the middle instruction.
7. The method for autonomously switching between water and land working conditions of an amphibious vehicle according to claim 6 is characterized by: The working condition conversion instruction is mapped to the actuator to realize the water-land working condition conversion. The working condition conversion instruction is mapped to the actuator specifically including: When the "onshore working condition" is switched to the "intermediate working condition", the water jet propulsion system is turned on, the wave-breaking board is deployed, and all the air intake devices are closed, and the transmission system switches from only land transmission to simultaneous water and land transmission; When the "intermediate working condition" is switched to the "on-water working condition", the sonar is turned on, the suspension is shortened, and the transmission system switches from working on both land and water to working only on water; When the "aquatic working condition" is switched to the "intermediate working condition", the sonar is turned off, the suspension is lowered, and the transmission system switches from working only on water to working on both land and water; When the "intermediate working condition" is switched to the "onland working condition", the water jet propulsion system is turned off, the wave-breaking board is retracted, the air intake device is opened, and the transmission system is switched from simultaneous water and land transmission to only land transmission operation.
Citation Information
Patent Citations
Unmanned vehicle behavior decision-making method based on ontology and Bayesian network
CN110968101A
Amphibious vehicle
CN113524996A
Amphibious multi-working-condition switching one-key control method for amphibious high-speed amphibious mobile platform
CN115743158A
Amphibious vehicle control method, system and device and storage medium
CN116512829A
Multi-power-source amphibious boat energy management control system and method
CN116736700A
Cited By
Working condition identification method for amphibious vehicle
CN118092156A
A method for identifying working conditions of an amphibious vehicle
CN118092156B
