Per-Contact Iteration Method for Solving Contact Dynamics

Hwangbo, Jemin; Lee, Joonho; Hutter, Marco

doi:10.1109/lra.2018.2792536

Cited by 177 publications

(151 citation statements)

References 17 publications

Supporting

Mentioning

150

Contrasting

Order By: Relevance

“…We expect computation time to grow quadratically with the number of contact points since the Delassus matrix (G) size depends quadratically on the contact points. This conjecture is confirmed by fitting a quadratic model to the recorded times and reproduces the result of state-of-the-art simulators [34]. Our method is hence well suited for tasks that require many contact interactions.…”

Section: B Scalability With Respect To the Number Of Contact Pointssupporting

confidence: 72%

Trajectory Optimization With Implicit Hard Contacts

Carius

Ranftl

Koltun

et al. 2018

IEEE Robot. Autom. Lett.

Self Cite

View full text Add to dashboard Cite

Section: B Scalability With Respect To the Number Of Contact Pointssupporting

confidence: 72%

Trajectory Optimization With Implicit Hard Contacts

Carius

Ranftl

Koltun

et al. 2018

IEEE Robot. Autom. Lett.

Self Cite

View full text Add to dashboard Cite

“…One of the biggest challenges with walking robots is the dynamics at intermittent contacts. To this end, we utilize the rigid body contact solver presented in our previous work [41]. This contact solver employs a hard contact model that fully respects the Coulomb friction cone constraint.…”

Section: Modeling Rigid-body Dynamicsmentioning

confidence: 99%

“…We used a fast custom implementation of the algorithm [55]. This efficient implementation and fast rigid-body simulation [41] allowed us to generate and process about a quarter of a billion state transitions in roughly four hours. A learning session terminates if the average performance of the policy does not improve by more than a task-specific threshold within 300 TRPO iterations.…”

Section: Reinforcement Learningmentioning

confidence: 99%

Learning agile and dynamic motor skills for legged robots

et al. 2019

Self Cite

View full text Add to dashboard Cite

Legged robots pose one of the greatest challenges in robotics. Dynamic and agile maneuvers of animals cannot be imitated by existing methods that are crafted by humans. A compelling alternative is reinforcement learning, which requires minimal craftsmanship and promotes the natural evolution of a control policy. However, so far, reinforcement learning research for legged robots is mainly limited to simulation, and only few and comparably simple examples have been deployed on real systems. The primary reason is that training with real robots, particularly with dynamically balancing systems, is complicated and expensive. In the present work, we report a new method for training a neural network policy in simulation and transferring it to a state-of-the-art legged system, thereby we leverage fast, automated, and cost-effective data generation schemes. The approach is applied to the ANYmal robot, a sophisticated medium-dog-sized quadrupedal system. 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 following high-level body velocity commands, running faster than ever before, and recovering from falling even in complex configurations.

show abstract

“…We realized the MDP environment for the GP using an own implementation of CROC in C++, while for the MDP environment of the GC we used the RaiSim [20] multi-body physics engine. All RL algorithms were implemented using the TensorFlow 3 C/C++ API.…”

Section: A Experimental Setupmentioning

confidence: 99%

DeepGait: Planning and Control of Quadrupedal Gaits Using Deep Reinforcement Learning

Tsounis

Alge

Lee

et al. 2020

IEEE Robot. Autom. Lett.

Self Cite

153

View full text Add to dashboard Cite

This paper addresses the problem of legged locomotion in non-flat terrain. As legged robots such as quadrupeds are to be deployed in terrains with geometries which are difficult to model and predict, the need arises to equip them with the capability to generalize well to unforeseen situations. In this work, we propose a novel technique for training neuralnetwork policies for terrain-aware locomotion, which combines state-of-the-art methods for model-based motion planning and reinforcement learning. Our approach is centered on formulating Markov decision processes using the evaluation of dynamic feasibility criteria in place of physical simulation. We thus employ policy-gradient methods to independently train policies which respectively plan and execute foothold and base motions in 3D environments using both proprioceptive and exteroceptive measurements. We apply our method within a challenging suite of simulated terrain scenarios which contain features such as narrow bridges, gaps and stepping-stones, and train policies which succeed in locomoting effectively in all cases. * These authors contributed equally.

show abstract

Per-Contact Iteration Method for Solving Contact Dynamics

Cited by 177 publications

References 17 publications

Trajectory Optimization With Implicit Hard Contacts

Trajectory Optimization With Implicit Hard Contacts

Learning agile and dynamic motor skills for legged robots

DeepGait: Planning and Control of Quadrupedal Gaits Using Deep Reinforcement Learning

Contact Info

Product

Resources

About