CN114819252A - New energy automobile unmanned path optimization method based on deep learning - Google Patents
New energy automobile unmanned path optimization method based on deep learning Download PDFInfo
- Publication number
- CN114819252A CN114819252A CN202210204941.1A CN202210204941A CN114819252A CN 114819252 A CN114819252 A CN 114819252A CN 202210204941 A CN202210204941 A CN 202210204941A CN 114819252 A CN114819252 A CN 114819252A
- Authority
- CN
- China
- Prior art keywords
- elm
- algorithm
- particle
- unmanned
- training
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000005457 optimization Methods 0.000 title claims abstract description 16
- 238000013135 deep learning Methods 0.000 title claims abstract description 10
- 239000002245 particle Substances 0.000 claims abstract description 50
- 238000012549 training Methods 0.000 claims abstract description 23
- 238000012360 testing method Methods 0.000 claims abstract description 22
- 210000002569 neuron Anatomy 0.000 claims abstract description 4
- 239000011159 matrix material Substances 0.000 claims description 12
- 238000010276 construction Methods 0.000 claims description 3
- 238000002790 cross-validation Methods 0.000 claims description 3
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000012545 processing Methods 0.000 description 2
- 230000003044 adaptive effect Effects 0.000 description 1
- 238000007792 addition Methods 0.000 description 1
- 230000002567 autonomic effect Effects 0.000 description 1
- 210000004027 cell Anatomy 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
- G06Q10/047—Optimisation of routes or paths, e.g. travelling salesman problem
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- Business, Economics & Management (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Human Resources & Organizations (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a new energy automobile unmanned path optimization method based on deep learning, which comprises the steps of firstly collecting parameters of unmanned path planning, classifying the collected parameters of the unmanned path planning as input of a deep ELM classifier; and constructing an ELM classifier: setting a proper number of hidden neurons at the stage, and training an unmanned path planning network model through an ELM algorithm; the RPSO algorithm optimizes the weight and offset of the ELM: selecting the optimal particles as the weight and the offset of an ELM algorithm through an RPSO algorithm, and classifying the test set through the model; and outputting the classification result to find the optimal route of the unmanned driving path planning. The invention carries out prediction through deep learning, and further obtains a more accurate prediction result.
Description
Technical Field
The invention belongs to the technical field of unmanned driving of new energy vehicles in the field of new energy vehicle batteries, and particularly relates to a method for optimizing unmanned driving paths of new energy vehicles based on deep learning.
Background
The unmanned automobile means that under the condition that people do not operate the automobile, the automobile senses the surrounding environment by virtue of sensing equipment borne by the automobile, and various running states of the automobile are controlled by a system program, so that the purpose of autonomous driving of the automobile is achieved. In unmanned automobile, the automobile body equips one set by system control's intelligent driving device, and this equipment can guarantee that the car carries out autonomic road conditions in the course of traveling and differentiates, when meetting dangerous condition, this equipment can be to the braking system of car and instruct, lets the car stop work to distance between the control vehicle that this equipment can also be fine in the course of traveling, can carry out the initiative to the obstacle on road surface and avoid, the car is at the whole operations of the in-process of traveling, independently be responsible for by this equipment. When the driving road condition is complex, the vehicle can combine the real-time peripheral condition according to the driving route which is set by the system in advance, and the independent intelligent operation is realized.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a new energy automobile unmanned path optimization method based on deep learning.
In order to achieve the purpose, the invention is realized by the following technical scheme:
a new energy automobile unmanned path optimization method based on deep learning comprises the following steps:
step 1: collecting parameters of the unmanned driving path planning, and classifying the collected parameters of the unmanned driving path planning as the input of a depth ELM classifier;
step 2: construction of ELM classifier: setting a proper number of hidden neurons at the stage, and training an unmanned path planning network model through an ELM algorithm;
and step 3: the RPSO algorithm optimizes the weight and offset of the ELM: selecting the optimal particles as the weight and the offset of an ELM algorithm through an RPSO algorithm, and classifying the test set through the model;
and 4, step 4: and outputting the classification result to find the optimal route of the unmanned driving path planning.
Further, the specific method for optimizing the weight and offset of the ELM by the RPSO algorithm in step 3 is as follows:
step 31: normalizing the original data samples collected in the step 1, and dividing the original data samples by a 5-fold cross-validation method, wherein 4 sample sets are used as training sets, and the rest sample sets are used as testing sets;
step 32: initializing parameters of RPSO and ELM algorithms, wherein the parameters comprise the number of particles in a population, the maximum iteration number of the population, a threshold (T0) of body history extreme value search stagnation, a global extreme value search stagnation threshold (Tg), learning factors c1 and c2 and a particle value range;
step 33: and randomly generating initial solutions with the value range of [ -1,1] and coding, wherein the number of the initial solutions is the number of particles in the population, the number of nodes of an input layer is N, the number of nodes of a hidden layer is L, the dimensionality of each solution is L (N +1) dimensionality, the former L multiplied by N dimensionality is input weight, and the rest L dimensionalities represent hidden layer node bias.
Step 34: decoding each solution containing the input weight and the hidden layer bias generated in the step 33 to construct an ELM model, and training the ELM by using a CV (model-constant-value) training sample; wherein, the solution obtained in step 33 is actually in the form of vector, but actually in the form of matrix in the process of ELM training (input weight matrix of dimension L × N and hidden layer bias of dimension L × 1), so the vector needs to be decoded into matrix form in order to construct the ELM model;
step 35: testing the ELM model trained in the step 34 by using test samples divided by a CV method to obtain a test result, and calculating a fitness value according to a formula;
step 36: recording a global extreme value and a particle individual extreme value of the particle swarm, judging whether the algorithm is in a stagnation state or not according to an algorithm stagnation step number judgment condition, if the algorithm is in the stagnation state, updating the motion speed and the position of the particle by using a formula, otherwise, updating the motion speed and the position of the particle by using a speed and position formula of a standard particle swarm optimization algorithm;
step 37: updating individual extreme values of the particles by adopting a greedy selection strategy, wherein the specific implementation mode is as follows: judging whether the fitness value of the current position (tempPos) of the particle is superior to the individual extreme value (pBest) of the position (tempPos) of the particle, if the tempPos is superior to the pBest, replacing the pBest with the tempPos, and if not, keeping the pBest unchanged;
step 38: updating the global extremum of the particle group by adopting a greedy selection strategy, judging whether the fitness value of the current position (tempPos) of the particle is superior to the global extremum (gBest) of the particle group, if the tempPos is superior to the gBest, replacing the gBest by the tempPos, and otherwise, keeping the gBest unchanged;
step 39: judging whether the particle swarm optimization ELM parameter algorithm process meets a set iteration termination condition (the maximum iteration number or the global fitness value meets a preset error epsilon), if so, turning to Setp10 for execution, otherwise, turning to the step 4 for continuing a new iteration execution;
step 310: decoding according to the returned optimal parameter combination to generate corresponding input weight and hidden layer bias;
step 311: training an ELM model by using the input weight returned in the step 10, hidden layer bias and a training sample set divided by a CV method;
step 312: inputting a test set divided by a CV method to the trained ELM model for classification test, and outputting a tested Root Mean Square Error (RMSE);
step 313: and judging whether the algorithm is operated for the 5 th time, if so, calculating the RMSE mean value of the five times, and storing the actual classification result, otherwise, returning to the step 2 to continue the parameter optimization and classification process for the next time.
Has the advantages that:
1. when the new energy automobile runs in an urban road, in the face of complex road conditions, the complex information increases the requirement of an unmanned path planning algorithm on a storage space, and the decision rate is reduced; when the vehicle encounters an obstacle, the vehicle can vibrate, swing can occur in a narrow channel, and the vehicle falls into a trap area, so that the vehicle is forced to stop and cannot reach a destination. According to the invention, the unmanned driving path planning parameters are continuously collected, and the BP-ELM-RPSO integrated algorithm is adopted, so that the driving path of the new energy automobile is planned, and the accuracy of the unmanned driving technology is further improved.
2. If the BP neural network contains hidden layers, continuous functions can be approximated at closed time intervals, which is the basic principle of BP network structural design. The invention solves the problem by increasing nodes in the hidden layer, so a method for increasing the number of the hidden layers cannot be selected, and the convergence performance of the network is related to the number of the hidden layers in the network. The use of multiple hidden layer BP neural networks therefore solves this problem and achieves good results, thereby improving the prediction accuracy of the network.
3. The invention combines an optimized group intelligence algorithm (RPSO) and an Extreme Learning Machine (ELM), uses the RPSO to optimize the ELM model input weight and the hidden layer bias, and verifies that the improved algorithm can obtain better generalization performance and classification accuracy rate than the ELM on the aspect of processing the classification problem. The basic idea of RPSO-ELM is: the ELM connection input weight and hidden layer bias are iteratively optimized through good random search performance of an adaptive extreme particle swarm algorithm, an ELM model is trained through training samples, a weight matrix connecting hidden layer nodes and output nodes is calculated through a Moore-Penrose generalized inverse matrix, and an ELM optimal parameter combination is found out so as to improve the classification effect of the ELM model.
Drawings
FIG. 1 is a flow chart of the RPSO-ELM algorithm of the present invention;
FIG. 2 is a comparison of the present invention with other deep neural network optimization algorithms.
Detailed Description
The invention is further described below with reference to the accompanying drawings.
As shown in fig. 1, the method for optimizing the unmanned driving path of the new energy vehicle based on deep learning comprises the following steps:
step 1: collecting parameters of the unmanned driving path planning, and classifying the collected parameters of the unmanned driving path planning as the input of a depth ELM classifier;
step 2: construction of ELM classifier: setting a proper number of hidden neurons at the stage, and training an unmanned path planning network model through an ELM algorithm;
and step 3: the RPSO algorithm optimizes the weight and offset of the ELM: selecting the optimal particles as the weight and the offset of an ELM algorithm through an RPSO algorithm, and classifying the test set through the model;
and 4, step 4: and outputting the classification result to find the optimal route of the unmanned driving path planning.
Further, the specific method for optimizing the weight and offset of the ELM by the RPSO algorithm in step 3 is as follows:
step 31: normalizing the original data samples collected in the step 1, and dividing the original data samples by a 5-fold cross-validation method, wherein 4 sample sets are used as training sets, and the rest sample sets are used as testing sets;
step 32: initializing parameters of RPSO and ELM algorithms, including the number of particles in the population, the maximum iteration number of the population, a threshold value (T0) of body history extreme value search stagnation, a global extreme value search stagnation threshold value (Tg), learning factors c1 and c2 and a particle value range;
step 33: and randomly generating initial solutions with the value range of [ -1,1] and coding, wherein the number of the initial solutions is the number of particles in the population, the number of nodes of an input layer is N, the number of nodes of a hidden layer is L, the dimensionality of each solution is L (N +1) dimensionality, the former L multiplied by N dimensionality is input weight, and the rest L dimensionalities represent hidden layer node bias.
Step 34: decoding each solution containing the input weight and the hidden layer bias generated in the step 33 to construct an ELM model, and training the ELM by using a CV (model-constant-value) training sample; wherein, the solution obtained in step 33 is actually in the form of vector, but actually in the form of matrix in the process of ELM training (input weight matrix of dimension L × N and hidden layer bias of dimension L × 1), so the vector needs to be decoded into matrix form in order to construct the ELM model;
step 35: testing the ELM model trained in the step 34 by using test samples divided by a CV method to obtain a test result, and calculating a fitness value according to a formula;
step 36: recording a global extreme value and a particle individual extreme value of the particle swarm, judging whether the algorithm is in a stagnation state according to the stagnation step number of the algorithm, if so, updating the motion speed and the position of the particle by using a formula, otherwise, updating the motion speed and the position of the particle by using a speed and position formula of a standard particle swarm optimization algorithm;
step 37: updating individual extreme values of the particles by adopting a greedy selection strategy, wherein the specific implementation mode is as follows: judging whether the fitness value of the current position (tempPos) of the particle is superior to the individual extreme value (pBest) of the position (tempPos) of the particle, if the tempPos is superior to the pBest, replacing the pBest with the tempPos, and if not, keeping the pBest unchanged;
step 38: updating the global extremum of the particle group by adopting a greedy selection strategy, judging whether the fitness value of the current position (tempPos) of the particle is superior to the global extremum (gBest) of the particle group, if the tempPos is superior to the gBest, replacing the gBest by the tempPos, and otherwise, keeping the gBest unchanged;
step 39: judging whether the particle swarm optimization ELM parameter algorithm process meets a set iteration termination condition (the maximum iteration number or the global fitness value meets a preset error epsilon), if so, turning to Setp10 for execution, otherwise, turning to the step 4 for continuing a new iteration execution;
step 310: decoding according to the returned optimal parameter combination to generate corresponding input weight and hidden layer bias;
step 311: training an ELM model by using the input weight returned in the step 10, hidden layer bias and a training sample set divided by a CV method;
step 312: inputting a test set divided by a CV method to the trained ELM model for classification test, and outputting a tested Root Mean Square Error (RMSE);
step 313: and judging whether the algorithm is operated for the 5 th time, if so, calculating the RMSE mean value of the five times, and storing the actual classification result, otherwise, returning to the step 2 to continue the parameter optimization and classification process for the next time.
In order to verify whether the improved algorithm is effectively applied to the path planning problem, a grid map method which is relatively direct to environment description and easy to represent and modify is adopted. In the virtual environment, a two-dimensional grid environment model with the size of 25 × 25 is established, the specification of all grid cells is 4m × 4m, see fig. 2 for details, a dotted line part represents an obstacle which is not subjected to expansion processing, and the remaining straight line part represents an obstacle-free object. As can be seen from FIG. 2, the algorithm provided by the patent can better plan the unmanned driving path, and compared with other algorithms, the accuracy and precision are greatly improved.
The specific embodiments described herein are merely illustrative of the spirit of the invention. Various modifications or additions may be made to the described embodiments or alternatives may be employed by those skilled in the art without departing from the spirit or ambit of the invention as defined in the appended claims.
Claims (2)
1. A new energy automobile unmanned path optimization method based on deep learning is characterized by comprising the following steps:
step 1: collecting parameters of the unmanned driving path planning, and classifying the collected parameters of the unmanned driving path planning as the input of a depth ELM classifier;
step 2: construction of ELM classifier: setting a proper number of hidden neurons at the stage, and training an unmanned path planning network model through an ELM algorithm;
and step 3: the RPSO algorithm optimizes the weight and offset of the ELM: selecting the optimal particles as the weight and the offset of an ELM algorithm through an RPSO algorithm, and classifying the test set through the model;
and 4, step 4: and outputting the classification result to find the optimal route of the unmanned driving path planning.
2. The method for optimizing the unmanned driving path of the new energy automobile based on deep learning of claim 1, wherein the specific method for optimizing the weight and offset of the ELM by the RPSO algorithm in step 3 is as follows:
step 31: normalizing the original data samples collected in the step 1, and dividing the original data samples by a 5-fold cross-validation method, wherein 4 sample sets are used as training sets, and the rest are used as test sets;
step 32: initializing parameters of an RPSO algorithm and an ELM algorithm, wherein the parameters comprise the number of particles in a population, the maximum iteration number of the population, a threshold T0 of body history extreme value search stagnation, a global extreme value search stagnation threshold Tg, learning factors c1 and c2 and a particle value range;
step 33: and randomly generating initial solutions with the value range of [ -1,1] and coding, wherein the number of the initial solutions is the number of particles in the population, the number of nodes of an input layer is N, the number of nodes of a hidden layer is L, the dimensionality of each solution is L (N +1) dimensionality, the former L multiplied by N dimensionality is input weight, and the rest L dimensionalities represent hidden layer node bias.
Step 34: decoding each solution containing the input weight and the hidden layer bias generated in the step 33 to construct an ELM model, and training the ELM by using a CV (model-constant-value) training sample; wherein, the solution obtained in step 33 is actually in the form of a vector, but actually is in the form of a matrix in the process of ELM training, and the matrix includes an input weight matrix of dimension L × N and hidden layer bias of dimension L × 1, so that the vector needs to be decoded into a matrix form in order to construct an ELM model;
step 35: testing the ELM model trained in the step 34 by using test samples divided by a CV method to obtain a test result, and calculating a fitness value according to a formula;
step 36: recording a global extreme value and a particle individual extreme value of the particle swarm, judging whether the algorithm is in a stagnation state according to the stagnation step number of the algorithm, if so, updating the motion speed and the position of the particle by using a formula, otherwise, updating the motion speed and the position of the particle by using a speed and position formula of a standard particle swarm optimization algorithm;
step 37: updating individual extreme values of the particles by adopting a greedy selection strategy, wherein the specific implementation mode is as follows: judging whether the fitness value of the tempPos at the current position of the particle is superior to an individual extreme value pBest of the position of the particle, if so, replacing the pBest with the tempPos, otherwise, keeping the pBest unchanged;
step 38: updating the global extremum of the particle group by adopting a greedy selection strategy, judging whether the fitness value of tempPos at the current position of the particle is superior to the global extremum gBest of the particle group, if the tempPos are superior to the gBest, replacing the gBest with the tempPos, and if not, keeping the gBest unchanged;
step 39: judging whether the particle swarm optimization ELM parameter algorithm process meets a set iteration termination condition, namely the maximum iteration times or the global fitness value meets a preset error epsilon, if so, turning to Setp10 for execution, otherwise, turning to step 4 for continuing a new iteration execution;
step 310: decoding according to the returned optimal parameter combination to generate corresponding input weight and hidden layer bias;
step 311: training an ELM model by using the input weight returned in the step 10, hidden layer bias and a training sample set divided by a CV method;
step 312: inputting a test set divided by a CV method to the trained ELM model for classification test, and outputting a tested Root Mean Square Error (RMSE);
step 313: and judging whether the algorithm is operated for the 5 th time, if so, calculating the RMSE mean value of the five times, and storing the actual classification result, otherwise, returning to the step 2 to continue the parameter optimization and classification process for the next time.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210204941.1A CN114819252A (en) | 2022-03-02 | 2022-03-02 | New energy automobile unmanned path optimization method based on deep learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210204941.1A CN114819252A (en) | 2022-03-02 | 2022-03-02 | New energy automobile unmanned path optimization method based on deep learning |
Publications (1)
Publication Number | Publication Date |
---|---|
CN114819252A true CN114819252A (en) | 2022-07-29 |
Family
ID=82529681
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210204941.1A Pending CN114819252A (en) | 2022-03-02 | 2022-03-02 | New energy automobile unmanned path optimization method based on deep learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN114819252A (en) |
Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109635914A (en) * | 2018-12-17 | 2019-04-16 | 杭州电子科技大学 | Optimization extreme learning machine trajectory predictions method based on hybrid intelligent Genetic Particle Swarm |
US20210374549A1 (en) * | 2020-05-29 | 2021-12-02 | Robert Bosch Gmbh | Meta-learned, evolution strategy black box optimization classifiers |
-
2022
- 2022-03-02 CN CN202210204941.1A patent/CN114819252A/en active Pending
Patent Citations (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109635914A (en) * | 2018-12-17 | 2019-04-16 | 杭州电子科技大学 | Optimization extreme learning machine trajectory predictions method based on hybrid intelligent Genetic Particle Swarm |
US20210374549A1 (en) * | 2020-05-29 | 2021-12-02 | Robert Bosch Gmbh | Meta-learned, evolution strategy black box optimization classifiers |
Non-Patent Citations (1)
Title |
---|
田连晓: "基于粒子群优化极限学习机的图像分割", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108803321B (en) | Autonomous underwater vehicle track tracking control method based on deep reinforcement learning | |
CN111142522B (en) | Method for controlling agent of hierarchical reinforcement learning | |
Zhang et al. | Mobile robot path planning based on improved localized particle swarm optimization | |
Huang et al. | Personalized trajectory planning and control of lane-change maneuvers for autonomous driving | |
Zhao et al. | A novel direct trajectory planning approach based on generative adversarial networks and rapidly-exploring random tree | |
Grigorescu et al. | Neurotrajectory: A neuroevolutionary approach to local state trajectory learning for autonomous vehicles | |
CN110749333A (en) | Unmanned vehicle motion planning method based on multi-objective optimization | |
CN106371439B (en) | Unified automatic driving transverse planning method and system | |
CN112389436B (en) | Safety automatic driving track changing planning method based on improved LSTM neural network | |
Rupprecht et al. | A survey for deep reinforcement learning in markovian cyber–physical systems: Common problems and solutions | |
Song et al. | Personalized car-following control based on a hybrid of reinforcement learning and supervised learning | |
Wang et al. | Multi-objective trajectory planning method based on the improved elitist non-dominated sorting genetic algorithm | |
CN114545921B (en) | Unmanned vehicle path planning algorithm based on improved RRT algorithm | |
CN116734877A (en) | Robot dynamic obstacle avoidance method based on improved A-algorithm and dynamic window method | |
CN113479187B (en) | Layered different-step-length energy management method for plug-in hybrid electric vehicle | |
Yuhang et al. | An Adaptive evolutionary multi-objective estimation of distribution algorithm and its application to multi-UAV path planning | |
Trasnea et al. | GridSim: a vehicle kinematics engine for deep neuroevolutionary control in autonomous driving | |
CN114819252A (en) | New energy automobile unmanned path optimization method based on deep learning | |
Smit et al. | Informed sampling-based trajectory planner for automated driving in dynamic urban environments | |
Regier et al. | Improving navigation with the social force model by learning a neural network controller in pedestrian crowds | |
Elallid et al. | Vehicles control: Collision avoidance using federated deep reinforcement learning | |
CN114537435B (en) | Real-time whole vehicle track planning method in automatic driving | |
CN114167860B (en) | Automatic driving optimal track generation method and device | |
Ojha et al. | An analysis of artificial immune system and genetic algorithm in urban path planning | |
CN114228748A (en) | Human-like automatic driving track planning method based on geometric path generation |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220729 |
|
RJ01 | Rejection of invention patent application after publication |