Map-based Multi-Policy Reinforcement Learning: Enhancing Adaptability of Robots by Deep Reinforcement Learning
Explore this paper's citation graph
Summary
Map-based Multi-Policy Reinforcement Learning (MMPRL), which aims to search and store multiple policies that encode different behavioral features while maximizing the expected reward in advance of the environment change, enables robots to quickly adapt to large changes without requiring any prior knowledge on the type of injuries that could occur.
- Type
- preprint
- Published
- 2017-10-17
- Cited by
- 12
- References
- 38
- Access
- Open access
- OpenAlex
- https://openalex.org/W2765397130
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:27378896
Keywords
Reinforcement learning, Adaptability, Computer science, Artificial intelligence, Robot
References
- Robots that can adapt like animals
- Playing Atari with Deep Reinforcement Learning
- Bayesian Approach to Global Optimization: Theory and Applications
- Resilient Machines Through Continuous Self-Modeling
- A review of evolutionary artificial neural networks
- Multiple chaotic central pattern generators with learning for legged locomotion and malfunction compensation
- An evolutionary algorithm that constructs recurrent neural networks
- Human-level control through deep reinforcement learning
- MuJoCo: A physics engine for model-based control
- Memory-based control with recurrent neural networks
- Abstract of: "Fast Damage Recovery in Robotics with the T-Resilience Algorithm"
- Deep Reinforcement Learning for Dialogue Generation
- Supervision via competition: Robot adversaries for learning tasks
- Reset-free Trial-and-Error Learning for Data-Efficient Robot Damage Recovery
- Deep Reinforcement Learning: An Overview
- FeUdal Networks for Hierarchical Reinforcement Learning
- Evolution Strategies as a Scalable Alternative to Reinforcement Learning
- Domain randomization for transferring deep neural networks from simulation to the real world
- Reset-free Trial-and-Error Learning for Robot Damage Recovery
- Autoencoder-augmented neuroevolution for visual doom playing
Cited by
- Discovering the elite hypervolume by leveraging interspecies correlation
- RETRACTED: Research on breakthrough and innovation of UAV mission planning method based on cloud computing-based reinforcement learning algorithm
- Deep Reinforcement Learning for Single-Shot Diagnosis and Adaptation in Damaged Robots
- Scaling MAP-Elites to deep neuroevolution
- Policy gradient assisted MAP-Elites
- Trends in the Control of Hexapod Robots: A Survey
- Telematics and Computing: 9th International Congress, WITCOM 2020, Puerto Vallarta, Mexico, November 2–6, 2020, Proceedings
- Efficient Quality-Diversity Optimization through Diverse Quality Species
- Reinforcement Learning with Adaptive Curriculum Dynamics Randomization for Fault-Tolerant Robot Control
- Latent-Conditioned Policy Gradient for Multi-Objective Deep Reinforcement Learning
- CoMoCAVs: Cohesive Decision-Guided Motion Planning for Connected and Autonomous Vehicles with Multi-Policy Reinforcement Learning
- Reinforcement Learning Applied to Hexapod Robot Locomotion: An Overview
Related papers
- Review of Deep Reinforcement Learning for Robot Manipulation
- A reinforcement learning with adaptive state space recruitment strategy for real autonomous mobile robots
- Adaptive Modular Reinforcement Learning for Robot Controlled in Multiple Environments
- Robot path planning based on deep reinforcement learning
- Robot path planning in complex environment based on delayed-optimization reinforcement learning
- Research on Reinforcement Learning Algorithm for Path Planning of Multiple Mobile Robots
- Adaptive Coverage Path Planning Policy for a Cleaning Robot with Deep Reinforcement Learning
- REINFORCEMENT LEARNING BASED OBSTACLE AVOIDANCE LEARNING FOR INTELLIGENT ROBOT
- A reinforcement-learning approach to robot navigation
- Reinforcement Learning based Method for Autonomous Navigation of Mobile Robots in Unknown Environments