Transferring Dexterous Manipulation from GPU Simulation to a Remote Real-World TriFinger

Allshire, Arthur; Mittal, Mayank; Lodaya, Varun; Makoviychuk, Viktor; Makoviichuk, Denys; Widmaier, Felix; Wüthrich, Manuel; Bauer, Sebastian; Handa, Ankur; Garg, Animesh

doi:10.48550/arxiv.2108.09779

Cited by 7 publications

(15 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…However, this approach was sensitive to α. Inspired by [2] the distance as ||k n − k f ||, where k n and k f are both tensors of 2-4 keypoints distributed along the nut central axis and endeffector approach axis, respectively. Intuitively, this method computes distance on a single manifold, obviating tuning.…”

Section: B Subpolicy: Pickmentioning

confidence: 99%

“…Quantitatively and through numerous visual comparisons, our physics simulation module enabled accurate, efficient, and robust simulation of contact-rich interactions of assets with real-world geometries and material properties. Furthermore, our module was built on PhysX, which has been evaluated under challenging sim-to-real conditions [2,18,81].…”

Section: F Contact Forcesmentioning

confidence: 99%

“…research has demonstrated a host of other applications: creating training environments for virtual robots [20,50,91,105], generating large-scale grasping datasets [21,33], inferring real-world material parameters [70,71], simulating tactile sensors [73,84,100], and training reinforcement learning (RL) agents for manipulation and locomotion [13,78]. Compelling works have now shown that RL policies trained in simulation can be transferred to the real world [2,4,12,27,62,81,103].…”

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Factory: Fast Contact for Robotic Assembly

Narang¹,

Storey²,

Akinola³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Robotic assembly is one of the oldest and most challenging applications of robotics. In other areas of robotics, such as perception and grasping, simulation has rapidly accelerated research progress, particularly when combined with modern deep learning. However, accurately, efficiently, and robustly simulating the range of contact-rich interactions in assembly remains a longstanding challenge. In this work, we present Factory, a set of physics simulation methods and robot learning tools for such applications. We achieve real-time or faster simulation of a wide range of contact-rich scenes, including simultaneous simulation of 1000 nut-and-bolt interactions. We provide 60 carefullydesigned part models, 3 robotic assembly environments, and 7 robot controllers for training and testing virtual robots. Finally, we train and evaluate proof-of-concept reinforcement learning policies for nut-and-bolt assembly. We aim for Factory to open the doors to using simulation for robotic assembly, as well as many other contact-rich applications in robotics. Please see our website for supplementary content, including videos. 1

show abstract

Section: B Subpolicy: Pickmentioning

confidence: 99%

Section: F Contact Forcesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Factory: Fast Contact for Robotic Assembly

Narang¹,

Storey²,

Akinola³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

“…where each c r is coordinate of a corner of a 3D bounding box around the end effector at a given time-step, and is a function of end effector position p r quaternion q r in global frame. This pose representation is the same as that in the loss function by Allshire et al [1]. We chose this pose representation because it casts rotation in position space, which mitigates the problem of combining heterogenous terms in the same function.…”

Section: Formulation In Roboticsmentioning

confidence: 99%

Assistive Tele-op: Leveraging Transformers to Collect Robotic Task Demonstrations

Clever¹,

Handa²,

Mazhar³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Sharing autonomy between robots and human operators could facilitate data collection of robotic task demonstrations to continuously improve learned models. Yet, the means to communicate intent and reason about the future are disparate between humans and robots. We present Assistive Tele-op, a virtual reality (VR) system for collecting robot task demonstrations that displays an autonomous trajectory forecast to communicate the robot's intent. As the robot moves, the user can switch between autonomous and manual control when desired. This allows users to collect task demonstrations with both a high success rate and with greater ease than manual teleoperation systems. Our system is powered by transformers, which can provide a window of potential states and actions far into the future -with almost no added computation time. A key insight is that human intent can be injected at any location within the transformer sequence if the user decides that the model-predicted actions are inappropriate. At every time step, the user can (1) do nothing and allow autonomous operation to continue while observing the robot's future plan sequence, or (2) take over and momentarily prescribe a different set of actions to nudge the model back on track. We host the videos and other supplementary material at https://sites.google.com/view/assistive-teleop.

show abstract

“…To ensure meaningful comparisons, we must ensure that our learning method not only improves sampleefficiency, but also wall-clock time. GPU-based physics simulation has shown remarkable effectiveness for accelerating model-free RL algorithms (Liang et al, 2018;Allshire et al, 2021), given this, we develop a GPU-based differentiable simulator that can compute gradients of standard robotics models over many environments in parallel. Our PyTorch-based simulator allows us to connect high-quality simulation with existing algorithms and tools.…”

Section: Introductionmentioning

confidence: 99%

Accelerated Policy Learning with Parallel Differentiable Simulation

Xu¹,

Makoviychuk²,

Narang³

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Deep reinforcement learning can generate complex control policies, but requires large amounts of training data to work effectively. Recent work has attempted to address this issue by leveraging differentiable simulators. However, inherent problems such as local minima and exploding/vanishing numerical gradients prevent these methods from being generally applied to control tasks with complex contact-rich dynamics, such as humanoid locomotion in classical RL benchmarks. In this work we present a high-performance differentiable simulator and a new policy learning algorithm (SHAC) that can effectively leverage simulation gradients, even in the presence of non-smoothness. Our learning algorithm alleviates problems with local minima through a smooth critic function, avoids vanishing/exploding gradients through a truncated learning window, and allows many physical environments to be run in parallel. We evaluate our method on classical RL control tasks, and show substantial improvements in sample efficiency and wall-clock time over state-of-the-art RL and differentiable simulation-based algorithms. In addition, we demonstrate the scalability of our method by applying it to the challenging high-dimensional problem of muscle-actuated locomotion with a large action space, achieving a greater than 17× reduction in training time over the best-performing established RL algorithm. More visual results are provided at: https://short-horizon-actor-critic.github.io/.

show abstract

Transferring Dexterous Manipulation from GPU Simulation to a Remote Real-World TriFinger

Cited by 7 publications

References 14 publications

Factory: Fast Contact for Robotic Assembly

Factory: Fast Contact for Robotic Assembly

Assistive Tele-op: Leveraging Transformers to Collect Robotic Task Demonstrations

Accelerated Policy Learning with Parallel Differentiable Simulation

Contact Info

Product

Resources

About