520 Hits in 3.2 sec

Learning Torque Control for Quadrupedal Locomotion [article]

Shuxiao Chen, Bike Zhang, Mark W. Mueller, Akshara Rai, Koushil Sreenath
2022 arXiv   pre-print
Reinforcement learning (RL) is a promising tool for developing controllers for quadrupedal locomotion.  ...  In this paper, we introduce a learning torque control framework for quadrupedal locomotion, which trains an RL policy that directly predicts joint torques at a high frequency, thus circumventing the use  ...  We would also like to thank Tingnan Zhang, Xingye Da and Jonathan Li for their insightful discussions on RL-based locomotion.  ... 
arXiv:2203.05194v1 fatcat:zqk7mmhhfvf2hcrueskhrse2je

Circus ANYmal: A Quadruped Learning Dexterous Manipulation with Its Limbs [article]

Fan Shi, Timon Homberger, Joonho Lee, Takahiro Miki, Moju Zhao, Farbod Farshidian, Kei Okada, Masayuki Inaba, Marco Hutter
2020 arXiv   pre-print
Quadrupedal robots are skillful at locomotion tasks while lacking manipulation skills, not to mention dexterous manipulation abilities.  ...  Inspired by the animal behavior and the duality between multi-legged locomotion and multi-fingered manipulation, we showcase a circus ball challenge on a quadrupedal robot, ANYmal.  ...  Model-free: In quadrupedal locomotion tasks, model-based RL combines learned gait with a model-based whole-body controller [6] .  ... 
arXiv:2011.08811v2 fatcat:yx7ivxic7fao3g4sqpf2d6g3s4

Learning Vision-Guided Dynamic Locomotion Over Challenging Terrains [article]

Zhaocheng Liu, Fernando Acero, Zhibin Li
2021 arXiv   pre-print
This work presents a deep reinforcement learning approach that learns a robust Lidar-based perceptual locomotion policy in a partially observable environment using Proximal Policy Optimisation.  ...  Legged robots are becoming increasingly powerful and popular in recent years for their potential to bring the mobility of autonomous agents to the next level.  ...  The motivation for developing learning-based control policies for legged locomotion is multifaceted.  ... 
arXiv:2109.04322v1 fatcat:fybbstugt5dntla7acj3t4cxui

Efficient Learning of Control Policies for Robust Quadruped Bounding using Pretrained Neural Networks [article]

Anqiao Li, Zhicheng Wang, Jun Wu, Qiuguo Zhu
2021 arXiv   pre-print
Bounding is one of the important gaits in quadrupedal locomotion for negotiating obstacles.  ...  We proposed an efficient approach to learn robust bounding gaits by first pretraining the deep neural network (DNN) using data from a robot that used conventional model-based controllers.  ...  To generate multiple locomotion quadrupedal locomotion for negotiating obstacles.  ... 
arXiv:2011.00446v2 fatcat:3nilvmz2mjcy3n7swd5e7a7kfy

GLiDE: Generalizable Quadrupedal Locomotion in Diverse Environments with a Centroidal Model [article]

Zhaoming Xie, Xingye Da, Buck Babich, Animesh Garg, Michiel van de Panne
2022 arXiv   pre-print
Model-free reinforcement learning (RL) for legged locomotion commonly relies on a physics simulator that can accurately predict the behaviors of every degree of freedom of the robot.  ...  In contrast, approximate reduced-order models are commonly used for many model predictive control strategies.  ...  Deep Reinforcement Learning for Quadrupedal Robots Deep RL has become a viable approach for synthesizing control policies for quadrupedal robots.  ... 
arXiv:2104.09771v3 fatcat:grrtzghawrf4tjiw4xtzynmx2q

Fast and Efficient Locomotion via Learned Gait Transitions [article]

Yuxiang Yang, Tingnan Zhang, Erwin Coumans, Jie Tan, Byron Boots
2021 arXiv   pre-print
We focus on the problem of developing energy efficient controllers for quadrupedal robots. Animals can actively switch gaits at different speeds to lower their energy consumption.  ...  We show that the learned hierarchical controller consumes much less energy across a wide range of locomotion speed than baseline controllers.  ...  Recently, reinforcement learning became a popular approach to learn locomotion policies for legged robots [9, 25, 26] .  ... 
arXiv:2104.04644v3 fatcat:sgwjnneatjgnvavhkqmv6vi4ku

Guided Constrained Policy Optimization for Dynamic Quadrupedal Robot Locomotion [article]

Siddhant Gangapurwala, Alexander Mitchell, Ioannis Havoutis
2020 arXiv   pre-print
Deep reinforcement learning (RL) uses model-free techniques to optimize task-specific control policies.  ...  Despite having emerged as a promising approach for complex problems, RL is still hard to use reliably for real-world applications.  ...  Jemin Hwangbo from RSL, ETH Zurich for his valuable inputs. We would also like to thank Luigi Campanaro and Benoit Casseau for helping us with our experiments.  ... 
arXiv:2002.09676v1 fatcat:han4m6t4obg37exzjjjrji3nmq

Towards automatic discovery of agile gaits for quadrupedal robots

Christian Gehring, Stelian Coros, Marco Hutter, Michael Bloesch, Peter Fankhauser, Markus A. Hoepflinger, Roland Siegwart
2014 2014 IEEE International Conference on Robotics and Automation (ICRA)  
We use our method to implement a flying trot, a bound and a pronking gait for StarlETH, a fully autonomous, dog-sized quadrupedal robot. CONFIDENTIAL. Limited circulation. For review only.  ...  Inspired by the motor learning principles observed in nature, we use an optimization approach to automatically discover and fine-tune parameters for agile gaits.  ...  find optimal values for the various parameters of our locomotion controllers.  ... 
doi:10.1109/icra.2014.6907476 dblp:conf/icra/GehringCHBFHS14 fatcat:6vq2qkblhba2tlo3s7jxhvdwgi

Learning agile and dynamic motor skills for legged robots

Jemin Hwangbo, Joonho Lee, Alexey Dosovitskiy, Dario Bellicoso, Vassilios Tsounis, Vladlen Koltun, Marco Hutter
2019 Science Robotics  
A compelling alternative is reinforcement learning, which requires minimal craftsmanship and promotes the natural evolution of a control policy.  ...  Using policies trained in simulation, the quadrupedal machine achieves locomotion skills that go beyond what had been achieved with prior methods: ANYmal is capable of precisely and energy-efficiently  ...  We next compared the learned controller with the best-performing existing locomotion controller available for ANYmal (12) .  ... 
doi:10.1126/scirobotics.aau5872 pmid:33137755 fatcat:m7yzgjsd7bcy3mid7dijo2jds4

Learning Active Spine Behaviors for Dynamic and Efficient Locomotion in Quadruped Robots [article]

Shounak Bhattacharya, Abhik Singla, Abhimanyu, Dhaivat Dholakiya, Shalabh Bhatnagar, Bharadwaj Amrutur, Ashitava Ghosal, Shishir Kolathaya
2019 arXiv   pre-print
Fast quadrupedal locomotion with active spine is an extremely hard problem, and involves a complex coordination between the various degrees of freedom.  ...  With this learning framework, the robot reached a bounding speed of 2.1 m/s with a maximum Froude number of 2.  ...  REINFORCEMENT LEARNING BASED CONTROLLER In this section, we will outline the deep reinforcement learning framework used for learning spine based locomotion behaviours. A.  ... 
arXiv:1905.06077v2 fatcat:4pynjra2fbgpldgel3data2r3i

Quadrupedal Robots with Stiff and Compliant Actuation

C. David Remy, Marco Hutter, Mark Hoepflinger, Michael Bloesch, Christian Gehring, Roland Siegwart
2012 at - Automatisierungstechnik  
In the broader context of quadrupedal locomotion, this overview article introduces and compares two platforms that are similar in structure, size, and morphology, yet differ greatly in their concept of  ...  We show how this conceptual difference influences design and control of the robots, compare the hardware of the two systems, and show exemplary their advantages in different applications.  ...  Its small counterpart, the LittleDog robot also received widespread attention throughout the DARPA Learning Locomotion challenge [16; 17] in which a variety of teams demonstrated quadrupedal robotic  ... 
doi:10.1524/auto.2012.1042 fatcat:aqqbtcoqnjcidmzk2adfkr4m6e

Learning Semantics-Aware Locomotion Skills from Human Demonstration [article]

Yuxiang Yang, Xiangyun Meng, Wenhao Yu, Tingnan Zhang, Jie Tan, Byron Boots
2022 arXiv   pre-print
In this work, we present a framework that learns semantics-aware locomotion skills from perception for quadrupedal robots, such that the robot can traverse through complex offroad terrains with appropriate  ...  For maximum traversability, we pair the speed policy with a gait selector, which selects a robust locomotion gait for each forward speed.  ...  While a low base 0.4 0.6 0.8 1.0 1.2 1. 4 Low-level Convex MPC Controller The low-level convex MPC controller computes and applies torques for each actuated degree of freedom, given the locomotion skills  ... 
arXiv:2206.13631v1 fatcat:dkw2pwr57zd7rpmmudr3yfsqvy

An Adaptable Approach to Learn Realistic Legged Locomotion without Examples [article]

Daniel Ordonez-Apraez, Antonio Agudo, Francesc Moreno-Noguer, Mario Martin
2022 arXiv   pre-print
for a bipedal and a quadrupedal robot.  ...  Learning controllers that reproduce legged locomotion in nature has been a long-time goal in robotics and computer graphics.  ...  We present results with a bipedal and quadrupedal robot in simulation for the model-free control case and a reactive policy acting directly on joint torques.  ... 
arXiv:2110.14998v2 fatcat:sabtxqs73jbd7nzf6rn4rmwu6e

Real-Time Trajectory Adaptation for Quadrupedal Locomotion using Deep Reinforcement Learning

Siddhant Gangapurwala, Mathieu Geisert, Romeo Orsolino, Maurice Fallon, Ioannis Havoutis
2021 2021 IEEE International Conference on Robotics and Automation (ICRA)  
We train a policy using deep reinforcement learning (RL) to introduce additive deviations to a reference trajectory in order to generate a feedback-based trajectory tracking system for a quadrupedal robot  ...  We present a control architecture for real-time adaptation and tracking of trajectories generated using a terrain-aware trajectory optimization solver.  ...  Recent work on RL for quadrupedal locomotion [8] , [9] has shown great promise for development of robust and dynamic model-free data-driven control techniques which directly map sensory information  ... 
doi:10.1109/icra48506.2021.9561639 fatcat:ohxlvuhgnrc47oy4tsygvq5dna

Animal Gaits on Quadrupedal Robots Using Motion Matching and Model-Based Control

Dongho Kang, Simon Zimmermann, Stelian Coros
2021 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)  
We demonstrate the efficacy of our control framework on a variety of quadrupedal robots in simulation.  ...  In this paper, we explore the challenge of generating animal-like walking motions for legged robots.  ...  Quadrupedal Locomotion Control Several model-based control techniques have been vigorously studied and successfully demonstrated on various quadrupedal robot platforms.  ... 
doi:10.1109/iros51168.2021.9635838 fatcat:gp63kziz75gldcuriplvoxlinu
« Previous Showing results 1 — 15 out of 520 results