Rigid-Soft Interactive Learning for Robust Grasping

Abstract

Inspired by widely used soft fingers on grasping, we propose a method of rigid-soft interactive learning, aiming at reducing the time of data collection. In this paper, we classify the interaction categories into Rigid-Rigid, Rigid-Soft, Soft-Rigid according to the interaction surface between grippers and target objects. We find experimental evidence that the interaction types between grippers and target objects play an essential role in the learning methods. We use soft, stuffed toys for training, instead of everyday objects, to reduce the integration complexity and computational burden and exploit such rigid-soft interaction by changing the gripper fingers to the soft ones when dealing with rigid, daily-life items such as the Yale-CMU-Berkeley (YCB) objects. With a small data collection of 5K picking attempts in total, our results suggest that such Rigid-Soft and Soft-Rigid interactions are transferable. Moreover, the combination of different grasp types shows better performance on the grasping test. We achieve the best grasping performance at 97.5% for easy YCB objects and 81.3% for difficult YCB objects while using a precise grasp with a twosoft-finger gripper to collect training data and power grasp with a four-soft-finger gripper to test.

Cocktail: Learn a Better Neural Network Controller from Multiple Experts via Adaptive Mixing and Robust Distillation

“…Machine learning techniques, particularly those based on neural networks, have seen rapidly growing applications in autonomous cyber-physical systems such as self-driving vehicles, smart buildings, and robotic systems. These learning-enabled cyber-physical systems (LE-CPSs) adopt machine learning techniques not only for perception of the environment [1], but increasingly also for control [2] and decision making, in large part due to their advantages in learning effective strategies without the need of developing complex, costly, and errorprone physical models [3]. However, applying neural networks for building autonomous CPSs still faces significant hurdles, particularly with concerns of their impact on system safety, robustness, and efficiency.…”

Flange-Based Hand-Eye Calibration Using a 3D Camera With High Resolution, Accuracy, and Frame Rate

“…While the ToF cameras are usually developed for consumer usage, which can produce point clouds in real-time frame rate with compromises in accuracy and resolution, structured light cameras exhibit the opposite characteristics for engineering automation. The 3D perception technologies have been widely applied to high-accuracy reconstruction (Chen et al, 2019 ), defect and surface inspection (Tang et al, 2019 ), and intelligent robot (Wan et al, 2020 ; Yang et al, 2020a ).…”

Learning-based Optoelectronically Innervated Tactile Finger for Rigid-Soft Interactive Grasping

“…This paper is a continuation of our earlier work with the omni-adaptive soft finger for rigid-soft interaction learning [17], finger configuration learning [18], and optical fiberbased grasp sensing [26]. In this paper, we propose a sensorized design of the omni-adaptive soft finger using multiple optical fibers embedded with friction enhanced soft surface.…”

“…Robotic fingers and grippers with a soft design have shown great potentials in grasping [15]. By leveraging material softness, they usually feature passive compliance [16], [17], [18] and underactuation [7], [19], [20], leading to a simple control during grasping. Inspired by the Fin-Ray Effect, [16] proposed a novel compliant lattice-structure fingers to enhance grasp robustness and has been widely used in [16], [21], [22].…”