Automated Coordination Strategy Design Using Genetic Programming for Dynamic Multipoint Dynamic Aggregation

Gao, Guanqiang; Mei, Yi; Xin, Bin; Jia, Ya-Hui; Browne, Will N.

doi:10.1109/tcyb.2021.3080044

Cited by 12 publications

(3 citation statements)

References 52 publications

(72 reference statements)

Supporting

Mentioning

Contrasting

Order By: Relevance

“…By mimicking human stride, biped robots can achieve greater mobility and adaptability, making them suitable for a variety of applications, such as humanoid companionship, search and rescue, and prosthetics [11]. However, establishing stable and robust bipedal walking remains challenging due to the complexity of the underlying dynamics and control processes [12].…”

Section: Introductionmentioning

confidence: 99%

Design Approach for Evolutionary Techniques Using Genetic Algorithms: Application for a Biped Robot to Learn to Walk and Rise after a Fall

Szabó

2023

Mathematics

View full text Add to dashboard Cite

In this research, a biped robot is programmed to learn to walk using genetic algorithms. The biped robot has the ability to rise after a fall. The system is made according to human genetics, where a child tries to learn to walk, takes a few steps, falls, and then stands up somehow after a fall. After this process, the system restarts and tries to walk again. The biped robot evolves after each generation, and after each generation it will have a more natural walking posture. The system is unpredictable, because the biped robot sometimes evolves faster and sometimes evolves slower, like in human genetics where some children are faster and other children are slower to walk. The biped robot is considered to have evolved enough after having a walking posture similar to that of a human and does not fall.

show abstract

Section: Introductionmentioning

confidence: 99%

Design Approach for Evolutionary Techniques Using Genetic Algorithms: Application for a Biped Robot to Learn to Walk and Rise after a Fall

Szabó

2023

Mathematics

View full text Add to dashboard Cite

show abstract

“…The Genetic Programming Hyper-heuristic (GPHH) is considered as a promising technique that can automatically evolve effective heuristics for dynamic problems [14], such as scheduling problems [15]- [18] and resource allocation problems [19]. Recently, GPHH has also been successfully applied to UCARP to evolve effective routing policies automatically [20]- [24].…”

mentioning

confidence: 99%

Genetic Programming With Niching for Uncertain Capacitated Arc Routing Problem

Wang

Mei

Zhang

et al. 2022

IEEE Trans. Evol. Computat.

Self Cite

View full text Add to dashboard Cite

The uncertain capacitated arc routing problem is an important optimisation problem with many real-world applications. Genetic programming is considered a promising hyperheuristic technique to automatically evolve routing policies that can make effective real-time decisions in an uncertain environment. Most existing research on genetic programming hyperheuristic for the uncertain capacitated arc routing problem only focused on the test performance aspect. As a result, the routing policies evolved by genetic programming are usually too large and complex, and hard to comprehend. To evolve effective, smaller and simpler routing policies, this paper proposes a novel genetic programming approach, which simplifies the routing policies during the evolutionary process using a niching technique. The simplified routing policies are stored in an external archive. We also developed new elitism, parent selection and breeding schemes for generating offspring from the original population and the archive. The experimental results show that the newly proposed approach can achieve significantly better test performance than the current state-of-the-art genetic programming algorithms for uncertain capacitated arc routing problem. The evolved routing policies are smaller, and thus potentially more interpretable.

show abstract

“…Recently, Genetic Programming Hyper Heuristic (GPHH) has been considered a promising technique that can automatically evolve effective heuristics for dynamic problems [1], such as scheduling problems [26,152,214,215] and resource allocation problems [180]. GPHH has also been successfully applied to UCARP to evolve effective routing policies automatically [67,122,123,133,143]. Although the GP-evolved routing policies have shown effectiveness, they are often too complex to be understood by real users.…”

Section: Problem Statementmentioning

confidence: 99%

Explainable Genetic Programming for Evolving Routing Policies of Uncertain Capacitated Arc Routing Problems

Wang

View full text Add to dashboard Cite

The Uncertain Capacitated Arc Routing Problem (UCARP) is a well-known combinatorial optimization problem that has many real-world applications. Genetic programming has successfully evolved routing policies that can make real-time routing decisions for uncertain arc routing problems. Although the evolved routing policies are highly effective, they are typically very complex thus hard for real users to understand and trust. Therefore, it is necessary to improve the interpretability of GP-evolved routing policies.The overall goal of this thesis is to develop both intrinsic and post-hoc methods to improve the interpretability of GP-evolved routing policies for UCARP.First, this thesis proposes a novel intrinsic genetic programming approach that simplifies the routing policies during the evolutionary process using a niching technique. The simplified routing policies are stored in an external archive. This thesis also develops new elitism, parent selection, and breeding schemes for generating offspring from the original population and the archive. The experimental results show that the newly proposed approach can achieve significantly better test effectiveness than the current state-of-the-art genetic programming hyper heuristic methods for UCARP. The evolved routing policies are smaller, thus potentially more interpretable.Second, this thesis develops several multi-objective genetic programming algorithms to evolve a Pareto front of routing policies with different interpretability and effectiveness, so that the end users can choose based on their preferences. This thesis handles two main challenges, i.e., objective selection bias issue and stochastic fitness evaluation issue, of designing Multi-Objective GP (MOGP) for UCARP. To handle the objective selection bias issue, an $\alpha$ dominance strategy is proposed. To further improve the $\alpha$ dominance strategy, this thesis develops a new $\alpha$ value adjustment scheme. To handle the stochastic fitness evaluation issue, this thesis proposes an archive strategy. The new archive strategy uses an external archive to store the potentially effective individuals that could have been lost during the traditional GP process. The individuals in the archive contribute back to the population in the breeding process. Finally, this thesis combines all the above strategies to develop a new MOGP algorithm. The experimental results show that the new MOGP algorithm outperforms state-of-the-art algorithms for UCARP in terms of HV and IGD. The evolved routing policies are much smaller, thus potentially more interpretable.Lastly, this thesis proposes a new post-hoc explanation approach to explaining the effective but complex GP-evolved routing policies. This thesis proposes two metrics that can evaluate the quality of a local explanation for GP-evolved routing policies. Then, this thesis investigates the feasibility of using a linear model to construct a local explanation for a complex GP-evolved routing policy in a decision situation. After that, this thesis further improves the local explanation method using Particle Swarm Optimization (PSO) to learn an interpretable linear model that accurately explains the local behavior of the routing policy for each decision situation, and extends the local explanation method to a global explanation method. The experimental results and case studies show that the proposed method can obtain accurate and understandable explanations of GP-evolved routing policies for UCARP.From the intrinsic aspect, both GP program simplification method and multi-objective GP methods are proposed. The GP program simplification approach addresses the case where only a single interpretable routing policy is required, i.e. where the user's preferences are known before the routing policy is generated. The multi-target approach is aimed at situations where several different interpretable routing policies are needed, because sometimes we cannot know the user's preferences in advance, and the user can choose their own routing policy according to their preferences. From the post-hoc aspect, this thesis proposes local and global explanation methods to further improve the interpretability of existing complex GP-evolved routing policies. Furthermore, the local and global explanation methods not only produce text explanations but also visual explanations.

show abstract

Automated Coordination Strategy Design Using Genetic Programming for Dynamic Multipoint Dynamic Aggregation

Cited by 12 publications

References 52 publications

Design Approach for Evolutionary Techniques Using Genetic Algorithms: Application for a Biped Robot to Learn to Walk and Rise after a Fall

Design Approach for Evolutionary Techniques Using Genetic Algorithms: Application for a Biped Robot to Learn to Walk and Rise after a Fall

Genetic Programming With Niching for Uncertain Capacitated Arc Routing Problem

Explainable Genetic Programming for Evolving Routing Policies of Uncertain Capacitated Arc Routing Problems

Contact Info

Product

Resources

About