Filters








86 Hits in 5.7 sec

Robust Biped Locomotion Using Deep Reinforcement Learning on Top of an Analytical Control Approach [article]

Mohammadreza Kasaei, Miguel Abreu, Nuno Lau, Artur Pereira, Luis Paulo Reis
2021 arXiv   pre-print
This paper proposes a modular framework to generate robust biped locomotion using a tight coupling between an analytical walking approach and deep reinforcement learning.  ...  This dynamics model is used to design an adaptive reference trajectories planner and an optimal controller which are fully parametric.  ...  Conclusion In this paper, we have tackled the problem of developing a robust biped locomotion framework by proposing a tight coupling between an analytical control approach and a reinforcement learning  ... 
arXiv:2104.10592v1 fatcat:ulkxmgotivdsrjdnysqimer7b4

Learning Hybrid Locomotion Skills – Learn to Exploit Residual Dynamics and Modulate Model-based Gait Control [article]

Mohammadreza Kasaei, Miguel Abreu, Nuno Lau, Artur Pereira, Luis Paulo Reis, Zhibin Li
2022 arXiv   pre-print
The framework embeds a kernel which is a fully parametric closed-loop gait generator based on analytical control.  ...  This work aims to combine machine learning and control approaches for legged robots, and developed a hybrid framework to achieve new capabilities of balancing against external perturbations.  ...  CONCLUSION In this paper, we proposed a locomotion framework based on a tight coupling between analytical control and deep reinforcement learning to combine the potential of both approaches.  ... 
arXiv:2011.13798v3 fatcat:dgrinscvufetbfmkyey66zbfnm

Learning locomotion skills using DeepRL

Xue Bin Peng, Michiel van de Panne
2017 Proceedings of the ACM SIGGRAPH / Eurographics Symposium on Computer Animation - SCA '17  
The use of deep reinforcement learning allows for high-dimensional state descriptors, but little is known about how the choice of action representation impacts the learning difficulty and the resulting  ...  We compare the impact of four different action parameterizations (torques, muscle-activations, target joint angles, and target joint-angle velocities) in terms of learning time, policy robustness, motion  ...  INTRODUCTION The introduction of deep learning models to reinforcement learning (RL) has enabled policies to operate directly on high-dimensional, low-level state features.  ... 
doi:10.1145/3099564.3099567 dblp:conf/sca/PengP17 fatcat:crcfk4lctzbatotjsejrkt2qle

Deep Reinforcement Learning for Tensegrity Robot Locomotion [article]

Marvin Zhang, Xinyang Geng, Jonathan Bruce, Ken Caluwaerts, Massimo Vespignani, Vytas SunSpiral, Pieter Abbeel, Sergey Levine
2017 arXiv   pre-print
the effectiveness of our approach on tensegrity robot locomotion.  ...  In this work, we show how locomotion gaits can be learned automatically using a novel extension of mirror descent guided policy search (MDGPS) applied to periodic locomotion movements, and we demonstrate  ...  Acknowledgments: This work was supported in part by the DARPA SIMPLEX program and an NSF CAREER award.  ... 
arXiv:1609.09049v3 fatcat:3xpdqdabgjbx5krke35kbxpnxu

Robust Dynamic Locomotion via Reinforcement Learning and Novel Whole Body Controller [article]

Donghyun Kim, Jaemin Lee, Luis Sentis
2017 arXiv   pre-print
Our locomotion strategy relies on devising a reinforcement learning (RL) approach for robust walking.  ...  We propose a robust dynamic walking controller consisting of a dynamic locomotion planner, a reinforcement learning process for robustness, and a novel whole-body locomotion controller (WBLC).  ...  ACKNOWLEDGMENT The authors would like to thank the members of the Human Centered Robotics Laboratory at The University of Texas at Austin for their great help and support.  ... 
arXiv:1708.02205v1 fatcat:w3ltjoku6fg6nkul35263iq3o4

Learning Locomotion Controllers for Walking Using Deep FBSDE [article]

Bolun Dai, Virinchi Roy Surabhi, Prashanth Krishnamurthy, Farshad Khorrami
2021 arXiv   pre-print
The efficacy of our approach is shown on a linear inverted pendulum model (LIPM) for walking.  ...  In this paper, we propose a deep forward-backward stochastic differential equation (FBSDE) based control algorithm for locomotion tasks.  ...  Recently, deep reinforcement learning has been widely applied to locomotion tasks.  ... 
arXiv:2107.07931v1 fatcat:lnacb2pjmjeunblqc7hbpvx6wq

Deep Reinforcement Learning for Industrial Insertion Tasks with Visual Inputs and Natural Rewards [article]

Gerrit Schoettler, Ashvin Nair, Jianlan Luo, Shikhar Bahl, Juan Aparicio Ojea, Eugen Solowjow, Sergey Levine
2019 arXiv   pre-print
Reinforcement learning (RL) methods have been demonstrated to be capable of learning controllers in such environments from autonomous interaction with the environment, but running RL algorithms in the  ...  We show that methods that combine RL with prior information, such as classical controllers or demonstrations, can solve these tasks from a reasonable amount of real-world interaction.  ...  To handle this challenge, we extend the residual RL approach [8, 9] , which learns a parametric policy on top of a fixed, hand-specified controller, to the setting of vision-based manipulation.  ... 
arXiv:1906.05841v2 fatcat:undptyk6h5gmtcazeyialzwzia

Detection of Landmines Using Nuclear Quadrupole Resonance (NQR): Signal Processing to Aid Classification [chapter]

S. D. Somasundaram, K. Althoefer, J. A. S. Smith, L. D. Seneviratne
2006 Climbing and Walking Robots  
These design drivers include: Lightweight Robustness Autonomy Controllable at high-level of abstraction Optimized locomotion Emphasis on collision avoidance These driving factors are described in more  ...  The paper deals with a pattern recognition approach to detect and classify falls of bipedal robots according to intensity and direction.  ...  Locomotion system is highly dependant on the environment nature. In some fields the most used solutions (wheels, legs or tracks) often show weakness.  ... 
doi:10.1007/3-540-26415-9_100 dblp:conf/clawar/SomasundaramASS05 fatcat:4vu25zlx65dszgbcxkusxozfve

Morphological Properties of Mass–Spring Networks for Optimal Locomotion Learning

Gabriel Urbain, Jonas Degrave, Benonie Carette, Joni Dambre, Francis Wyffels
2017 Frontiers in Neurorobotics  
On the one hand, the use of flexible materials can lead to higher power efficiency and more fluent and robust motions.  ...  We first investigate the influence of the network size and compliance on locomotion quality and energy efficiency by optimizing an external open-loop controller using evolutionary algorithms.  ...  The data were analyzed by GU with help of FW, JDegrave, and JDambre. The manuscript was mostly written by GU, with comments and corrections from FW and JDambre.  ... 
doi:10.3389/fnbot.2017.00016 pmid:28396634 pmcid:PMC5366341 fatcat:iz2jxwlxgfaxrcldq23gvilvvy

Deep Reinforcement Learning, a textbook [article]

Aske Plaat
2022 arXiv   pre-print
The way in which deep reinforcement learning explores complex environments reminds us of how children learn, by playfully trying out things, getting feedback, and trying again.  ...  The aim of this book is to provide a comprehensive overview of the field of deep reinforcement learning.  ...  What is Deep Reinforcement Learning? Deep reinforcement learning is the combination of deep learning and reinforcement learning.  ... 
arXiv:2201.02135v2 fatcat:3icsopexerfzxa3eblpu5oal64

Policy Transfer via Kinematic Domain Randomization and Adaptation [article]

Ioannis Exarchos, Yifeng Jiang, Wenhao Yu, C. Karen Liu
2021 arXiv   pre-print
Our algorithm, Multi-Policy Bayesian Optimization, trains an ensemble of universal policies conditioned on virtual kinematic parameters and efficiently adapts to the target environment using a limited  ...  Transferring reinforcement learning policies trained in physics simulation to the real hardware remains a challenge, known as the "sim-to-real" gap.  ...  INTRODUCTION The advent of Deep Reinforcement Learning (DRL) has demonstrated a promising approach to design robotic controllers for diverse robot motor skills [1] , [2] , [3] , [4] .  ... 
arXiv:2011.01891v3 fatcat:jmsbzzw4lzh43kisclw57vmvxm

Mapless Humanoid Navigation Using Learned Latent Dynamics [article]

Andre Brandenburger, Diego Rodriguez, Sven Behnke
2021 arXiv   pre-print
In this paper, we propose a novel Deep Reinforcement Learning approach to address the mapless navigation problem, in which the locomotion actions of a humanoid robot are taken online based on the knowledge  ...  In addition, we incorporate a termination likelihood predictor model as an auxiliary loss function of the control policy, which enables the agent to anticipate terminal states of success and failure.  ...  [21] investigate visual navigation on a bipedal platform and learned a control policy by using DDPG.  ... 
arXiv:2108.03866v1 fatcat:5ll7nbvus5hvvhunz2stgjmt4y

A posture sequence learning system for an anthropomorphic robotic hand

Antonio Chella, Haris Džindo, Ignazio Infantino, Irene Macaluso
2004 Robotics and Autonomous Systems  
Work in that area tackles the development of robust algorithms for motor control, motor learning, gesture recognition and visuo-motor integration.  ...  Studies of imitation in biological systems stressed a number of key issues that are, also, of concerns to Robotics.  ...  The work is being done with the support of CYCIT, Spanish Research Agency, under the projects: DPI2001-0822 and DPI2002-04286-C2-02 Acknowledgments: Among many people who contributed to the robot system  ... 
doi:10.1016/j.robot.2004.03.008 fatcat:fefnr5l7gnhnhaca5nw52f3kru

Safe Value Functions [article]

Pierre-François Massiani, Steve Heim, Friedrich Solowjow, Sebastian Trimpe
2021 arXiv   pre-print
Although these criteria are often solved separately with different tools to maintain formal guarantees, it is also common practice in reinforcement learning to simply modify reward functions by penalizing  ...  Safety constraints and optimality are important, but sometimes conflicting criteria for controllers.  ...  Levine, “Learning to walk via deep reinforcement “A learnable safety measure,” in Conference on Robot learning,” in Proceedings of Robotics: Science and Learning (CoRL), PMLR, 2019  ... 
arXiv:2105.12204v2 fatcat:5tc7aqflc5etljp3ceux55vmvy

DeepMind Control Suite [article]

Yuval Tassa, Yotam Doron, Alistair Muldal, Tom Erez, Yazhe Li, Diego de Las Casas, David Budden, Abbas Abdolmaleki, Josh Merel, Andrew Lefrancq, Timothy Lillicrap, Martin Riedmiller
2018 arXiv   pre-print
The DeepMind Control Suite is a set of continuous control tasks with a standardised structure and interpretable rewards, intended to serve as performance benchmarks for reinforcement learning agents.  ...  The tasks are written in Python and powered by the MuJoCo physics engine, making them easy to use and modify. We include benchmarks for several learning algorithms.  ...  However, continuous control domains cannot be solved by humans, so a different approach must be taken. In order to tackle both of these challenges, we ran variety of learning agents (e.g.  ... 
arXiv:1801.00690v1 fatcat:nsiln5qimje7pggv3ysh3k5swu
« Previous Showing results 1 — 15 out of 86 results