EfficientGrasp: A Unified Data-Efficient Learning to Grasp Method for Multi-Fingered Robot Hands

Li, Kelin; Baron, Nicholas; Zhang, Xian; Rojas, Nicolás

doi:10.1109/lra.2022.3187875

Cited by 21 publications

(9 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to evaluate the characteristic level of our PRH, a quantitative comparison of the response time, resolution, or accuracy, response hysteresis, and frequency feature of our PRH and other robotic hands is accomplished (Fig. 6b ) 43 – 49 . It can be found that the response time of the PRH is obviously short than other robotic hands due to the fast response capability of the functional piezoelectric ceramics.…”

Section: Discussionmentioning

confidence: 99%

Piezo robotic hand for motion manipulation from micro to macro

et al. 2023

View full text Add to dashboard Cite

Multiple degrees of freedom (DOFs) motion manipulation of various objects is a crucial skill for robotic systems, which relies on various robotic hands. However, traditional robotic hands suffer from problems of low manipulation accuracy, poor electromagnetic compatibility and complex system due to limitations in structures, principles and transmissions. Here we present a direct-drive rigid piezo robotic hand (PRH) constructed on functional piezoelectric ceramic. Our PRH holds four piezo fingers and twelve motion DOFs. It achieves high adaptability motion manipulation of ten objects employing pre-planned functionalized hand gestures, manipulating plates to achieve 2L (linear) and 1R (rotary) motions, cylindrical objects to generate 1L and 1R motions and spherical objects to produce 3R motions. It holds promising prospects in constructing multi-DOF ultra-precision manipulation devices, and an integrated system of our PRH is developed to implement several applications. This work provides a new direction to develop robotic hand for multi-DOF motion manipulation from micro scale to macro scale.

show abstract

Section: Discussionmentioning

confidence: 99%

Piezo robotic hand for motion manipulation from micro to macro

et al. 2023

View full text Add to dashboard Cite

show abstract

“…For example, traditional control methods were applied to robot arm systems performing machining task [17] and coffee machine tasks [2]. Dexterous manipulation skills, on the other hand, are typically achieved with learning-based methods, such as deep learning to grasp complex objects [18], or learning from human demonstration to perform unscrewing tasks [3], door opening [19], and in-hand manipulation [4].…”

Section: A Robotic Manipulationmentioning

confidence: 99%

Immersive Demonstrations are the Key to Imitation Learning

Li¹,

Chappell²,

Rojas³

2023

Preprint

View full text Add to dashboard Cite

Achieving successful robotic manipulation is an essential step towards robots being widely used in industry and home settings. Recently, many learning-based methods have been proposed to tackle this challenge, with imitation learning showing great promise. However, imperfect demonstrations and a lack of feedback from teleoperation systems may lead to poor or even unsafe results. In this work we explore the effect of demonstrator force feedback on imitation learning, using a feedback glove and a robot arm to render fingertip-level and palm-level forces, respectively. 10 participants recorded 5 demonstrations of a pick-and-place task with 3 grippers, under conditions with no force feedback, fingertip force feedback, and fingertip and palm force feedback. Results show that force feedback significantly reduces demonstrator fingertip and palm forces, leads to a lower variation in demonstrator forces, and recorded trajectories that a quicker to execute. Using behavioral cloning, we find that agents trained to imitate these trajectories mirror these benefits, even though agents have no force data shown to them during training. We conclude that immersive demonstrations, achieved with force feedback, may be the key to unlocking safer, quicker to execute dexterous manipulation policies.

show abstract

“…Other methods take an indirect approach that involves generating an intermediate representation first. Existing methods use contact points [47][48][49], contact maps [28,29,[50][51][52], and occupancy fields [53] as the intermediate representations.…”

Section: B Data-driven Graspingmentioning

confidence: 99%

“…The methods then obtain the grasping poses via optimization [47,48,51,53], planning [50], RL policies [29,49], or another generative model [52].…”

Section: B Data-driven Graspingmentioning

confidence: 99%

DexGraspNet: A Large-Scale Robotic Dexterous Grasp Dataset for General Objects Based on Simulation

Wang¹,

Zhang²,

Chen³

et al. 2022

Preprint

View full text Add to dashboard Cite

Object grasping using dexterous hands is a crucial yet challenging task for robotic dexterous manipulation. Compared with the field of object grasping with parallel grippers, dexterous grasping is very under-explored, partially owing to the lack of a large-scale dataset. In this work, we present a large-scale simulated dataset, DexGraspNet, for robotic dexterous grasping, along with a highly efficient synthesis method for diverse dexterous grasping synthesis. Leveraging a highly accelerated differentiable force closure estimator, we, for the first time, are able to synthesize stable and diverse grasps efficiently and robustly. We choose ShadowHand, a dexterous gripper commonly seen in robotics, and generated 1.32 million grasps for 5355 objects, covering more than 133 object categories and containing more than 200 diverse grasps for each object instance, with all grasps having been validated by the physics simulator. Compared to the previous dataset generated by GraspIt!, our dataset has not only more objects and grasps, but also higher diversity and quality. Via performing crossdataset experiments, we show that training several algorithms of dexterous grasp synthesis on our datasets significantly outperforms training on the previous one, demonstrating the large scale and diversity of DexGraspNet. We will release the data and tools upon acceptance.

show abstract

EfficientGrasp: A Unified Data-Efficient Learning to Grasp Method for Multi-Fingered Robot Hands

Cited by 21 publications

References 32 publications

Piezo robotic hand for motion manipulation from micro to macro

Piezo robotic hand for motion manipulation from micro to macro

Immersive Demonstrations are the Key to Imitation Learning

DexGraspNet: A Large-Scale Robotic Dexterous Grasp Dataset for General Objects Based on Simulation

Contact Info

Product

Resources

About