Abstract
Many real world reinforcement learning (RL) problems consist of multiple conflicting objective functions that need to be optimized simultaneously. Finding these optimal policies (known as Pareto optimal policies) for different preferences of objectives requires extensive state space exploration. Thus, obtaining a dense set of Pareto optimal policies is challenging and often reduces the sample efficiency. In this paper, we propose a hybrid multiobjective policy optimization approach for solving multiobjective reinforcement learning (MORL) problems with continuous actions. Our approach combines the faster convergence of multiobjective policy gradient (MOPG) and a surrogate assisted multiobjective evolutionary algorithm (MOEA) to produce a dense set of Pareto optimal policies. The solutions found by the MOPG algorithm are utilized to build computationally inexpensive surrogate models in the parameter space of the policies that approximate the return of policies. An MOEA is executed that utilizes the surrogates’ mean prediction and uncertainty in the prediction to find approximate optimal policies. The final solution policies are later evaluated using the simulator and stored in an archive. Tests on multiobjective continuous action RL benchmarks show that a hybrid surrogate assisted multiobjective evolutionary optimizer with robust selection criterion produces a dense set of Pareto optimal policies without extensively exploring the state space. We also apply the proposed approach to train Pareto optimal agents for autonomous driving, where the hybrid approach produced superior results compared to a state-of-the-art MOPG algorithm.
Notes
- 1.
Codes can be found at https://github.com/amrzr/SA-MOEAMOPG.
References
Ao, Y., Li, H., Zhu, L., Ali, S., Yang, Z.: The linear random forest algorithm and its advantages in machine learning assisted logging regression modeling. J. Petrol. Sci. Eng. 174, 776–789 (2019)
Arashi, M., Lukman, A.F., Algamal, Z.Y.: Liu regression after random forest for prediction and modeling in high dimension. J. Chemometr. 36(4), e3393 (2022)
Bouhlel, M.A., Martins, J.R.R.A.: Gradient-enhanced kriging for high-dimensional problems. Eng. Comput. 35(1), 157–173 (2018)
Chen, D., Wang, Y., Gao, W.: Combining a gradient-based method and an evolution strategy for multi-objective reinforcement learning. Appl. Intell. 50(10), 3301–3317 (2020)
Cheng, R., Jin, Y., Olhofer, M., Sendhoff, B.: A reference vector guided evolutionary algorithm for many-objective optimization. IEEE Trans. Evol. Comput. 20, 773–791 (2016)
Chugh, T., Sindhya, K., Hakanen, J., Miettinen, K.: A survey on handling computationally expensive multiobjective optimization problems with evolutionary algorithms. Soft. Comput. 23, 3137–3166 (2019)
Conlon, J., Lin, J.: Greenhouse gas emission impact of autonomous vehicle introduction in an urban network. Transp. Res. Rec. 2673(5), 142–152 (2019)
Deb, K., Jain, H.: An evolutionary many-objective optimization algorithm using reference-point-based nondominated sorting approach, part I: Solving problems with box constraints. IEEE Trans. Evol. Comput. 18, 577–601 (2014)
Deb, K., Pratap, A., Agarwal, S., Meyarivan, T.: A fast and elitist multiobjective genetic algorithm: NSGA-II. IEEE Trans. Evol. Comput. 6(2), 182–197 (2002)
Forrester, A., Sobester, A., Keane, A.: Engineering Design via Surrogate Modelling. John Wiley & Sons, Hoboken (2008)
Hayes, C.F., Reymond, M., Roijers, D.M., Howley, E., Mannion, P.: Risk aware and multi-objective decision making with distributional monte carlo tree search (2021). arXiv:2102.00966
Hayes, C.F., et al.: A practical guide to multi-objective reinforcement learning and planning. Auton. Agents Multi-Agent Syst. 36(1), 26 (2022)
Jin, Y.: Surrogate-assisted evolutionary computation: recent advances and future challenges. Swarm Evol. Comput. 1, 61–70 (2011)
Jin, Y., Wang, H., Chugh, T., Guo, D., Miettinen, K.: Data-driven evolutionary optimization: an overview and case studies. IEEE Trans. Evol. Comput. 23, 442–458 (2019)
Knowles, J.D., Thiele, L., Zitzler, E.: A tutorial on the performance assessment of stochastic multiobjective optimizers (2006)
Leurent, E.: An environment for autonomous driving decision-making (2018). https://github.com/eleurent/highway-env
Li, M., Yao, X.: Quality evaluation of solution sets in multiobjective optimisation. ACM Comput. Surv. 52(2), 1–38 (2019)
Mazumdar, A., Chugh, T., Hakanen, J., Miettinen, K.: Probabilistic selection approaches in decomposition-based evolutionary algorithms for offline data-driven multiobjective optimization. IEEE Trans. Evol. Comput. 26, 1182–1191 (2022)
Parisi, S., Pirotta, M., Smacchia, N., Bascetta, L., Restelli, M.: Policy gradient approaches for multi-objective sequential decision making. In: 2014 International Joint Conference on Neural Networks (IJCNN), pp. 2323–2330 (2014)
Rodriguez-Galiano, V., Sanchez-Castillo, M., Chica-Olmo, M., Chica-Rivas, M.: Machine learning predictive models for mineral prospectivity: an evaluation of neural networks, random forest, regression trees and support vector machines. Ore Geol. Rev. 71, 804–818 (2015)
Siddique, U., Weng, P., Zimmer, M.: Learning fair policies in multiobjective (deep) reinforcement learning with average and discounted rewards. In: Proceedings of the 37th International Conference on Machine Learning (2020)
Stork, J., et al.: Open issues in surrogate-assisted optimization. High-Performance Simulation-Based Optimization p. 225–244 (2019)
Xu, J., Tian, Y., Ma, P., Rus, D., Sueda, S., Matusik, W.: Prediction-guided multi-objective reinforcement learning for continuous robot control. In: Proceedings of the 37th International Conference on Machine Learning, pp. 10607–10616. PMLR (2020)
Yang, K., Emmerich, M., Deutz, A., Bäck, T.: Efficient computation of expected hypervolume improvement using box decomposition algorithms. J. Global Optim. 75(1), 3–34 (2019)
Zapotecas Martínez, S., Coello Coello, C.A.: Moea/d assisted by RBF networks for expensive multi-objective optimization problems. In: Proceedings of the 15th Annual Conference on Genetic and Evolutionary Computation, pp. 1405–1412. Association for Computing Machinery (2013)
Zhang, Q., Li, H.: MOEA/D: a multiobjective evolutionary algorithm based on decomposition. IEEE Trans. Evol. Comput. 11, 712–731 (2007)
Zitzler, E., Deb, K., Thiele, L.: Comparison of multiobjective evolutionary algorithms: empirical results. Evol. Comput. 8, 173–195 (2000)
Acknowledgements
The work is supported by Artificial Intelligence for Urban Low-Emission Autonomous Traffic (AIforLessAuto) funded under the Green and Digital transition call from the Academy of Finland. The research project has been granted funding from the European Union (NextGenerationEU) through the Academy of Finland under project number 347199.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2024 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Mazumdar, A., Kyrki, V. (2024). Hybrid Surrogate Assisted Evolutionary Multiobjective Reinforcement Learning for Continuous Robot Control. In: Smith, S., Correia, J., Cintrano, C. (eds) Applications of Evolutionary Computation. EvoApplications 2024. Lecture Notes in Computer Science, vol 14635. Springer, Cham. https://doi.org/10.1007/978-3-031-56855-8_4
Download citation
DOI: https://doi.org/10.1007/978-3-031-56855-8_4
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-56854-1
Online ISBN: 978-3-031-56855-8
eBook Packages: Computer ScienceComputer Science (R0)