SuctionPrompt: Visual-assisted Robotic Picking with a Suction Cup Using Vision-Language Models and Facile Hardware Design

The development of large language models and vision-language models (VLMs) has resulted in the increasing use of robotic systems in various fields. However, the effective integration of these models into real-world robotic tasks is a key challenge. We developed a versatile robotic system called SuctionPrompt that utilizes prompting techniques of VLMs combined with 3D detections to perform product-picking tasks in diverse and dynamic environments. Our method highlights the importance of integrating 3D spatial information with adaptive action planning to enable robots to approach and manipulate objects in novel environments. In the validation experiments, the system accurately selected suction points 75.4%, and achieved a 65.0% success rate in picking common items. This study highlights the effectiveness of VLMs in robotic manipulation tasks, even with simple 3D processing.

Visual Imitation Learning of Non-Prehensile Manipulation Tasks with Dynamics-Supervised Models

Unlike quasi-static robotic manipulation tasks like pick-and-place, dynamic tasks such as non-prehensile manipulation pose greater challenges, especially for vision-based control. Successful control requires the extraction of features relevant to the target task. In visual imitation learning settings, these features can be learnt by backpropagating the policy loss through the vision backbone. Yet, this approach tends to learn task-specific features with limited generalizability. Alternatively, learning world models can realize more generalizable vision backbones. Utilizing the learnt features, task-specific policies are subsequently trained. Commonly, these models are trained solely to predict the next RGB state from the current state and action taken. But only-RGB prediction might not fully-capture the task-relevant dynamics. In this work, we hypothesize that direct supervision of target dynamic states (Dynamics Mapping) can learn better dynamics-informed world models. Beside the next RGB reconstruction, the world model is also trained to directly predict position, velocity, and acceleration of environment rigid bodies. To verify our hypothesis, we designed a non-prehensile 2D environment tailored to two tasks: "Balance-Reaching" and "Bin-Dropping". When trained on the first task, dynamics mapping enhanced the task performance under different training configurations (Decoupled, Joint, End-to-End) and policy architectures (Feedforward, Recurrent). Notably, its most significant impact was for world model pretraining boosting the success rate from 21% to 85%. Although frozen dynamics-informed world models could generalize well to a task with in-domain dynamics, but poorly to a one with out-of-domain dynamics.

Component Selection for Craft Assembly Tasks

Inspired by traditional handmade crafts, where a person improvises assemblies based on the available objects, we formally introduce the Craft Assembly Task. It is a robotic assembly task that involves building an accurate representation of a given target object using the available objects, which do not directly correspond to its parts. In this work, we focus on selecting the subset of available objects for the final craft, when the given input is an RGB image of the target in the wild. We use a mask segmentation neural network to identify visible parts, followed by retrieving labelled template meshes. These meshes undergo pose optimization to determine the most suitable template. Then, we propose to simplify the parts of the transformed template mesh to primitive shapes like cuboids or cylinders. Finally, we design a search algorithm to find correspondences in the scene based on local and global proportions. We develop baselines for comparison that consider all possible combinations, and choose the highest scoring combination for common metrics used in foreground maps and mask accuracy. Our approach achieves comparable results to the baselines for two different scenes, and we show qualitative results for an implementation in a real-world scenario.

PEGASUS: Physically Enhanced Gaussian Splatting Simulation System for 6DOF Object Pose Dataset Generation

We introduce Physically Enhanced Gaussian Splatting Simulation System (PEGASUS) for 6DOF object pose dataset generation, a versatile dataset generator based on 3D Gaussian Splatting. Environment and object representations can be easily obtained using commodity cameras to reconstruct with Gaussian Splatting. PEGASUS allows the composition of new scenes by merging the respective underlying Gaussian Splatting point cloud of an environment with one or multiple objects. Leveraging a physics engine enables the simulation of natural object placement within a scene through interaction between meshes extracted for the objects and the environment. Consequently, an extensive amount of new scenes - static or dynamic - can be created by combining different environments and objects. By rendering scenes from various perspectives, diverse data points such as RGB images, depth maps, semantic masks, and 6DoF object poses can be extracted. Our study demonstrates that training on data generated by PEGASUS enables pose estimation networks to successfully transfer from synthetic data to real-world data. Moreover, we introduce the Ramen dataset, comprising 30 Japanese cup noodle items. This dataset includes spherical scans that captures images from both object hemisphere and the Gaussian Splatting reconstruction, making them compatible with PEGASUS.

NeuralLabeling: A versatile toolset for labeling vision datasets using Neural Radiance Fields

We present NeuralLabeling, a labeling approach and toolset for annotating a scene using either bounding boxes or meshes and generating segmentation masks, affordance maps, 2D bounding boxes, 3D bounding boxes, 6DOF object poses, depth maps and object meshes. NeuralLabeling uses Neural Radiance Fields (NeRF) as renderer, allowing labeling to be performed using 3D spatial tools while incorporating geometric clues such as occlusions, relying only on images captured from multiple viewpoints as input. To demonstrate the applicability of NeuralLabeling to a practical problem in robotics, we added ground truth depth maps to 30000 frames of transparent object RGB and noisy depth maps of glasses placed in a dishwasher captured using an RGBD sensor, yielding the Dishwasher30k dataset. We show that training a simple deep neural network with supervision using the annotated depth maps yields a higher reconstruction performance than training with the previously applied weakly supervised approach.

Motion Priority Optimization Framework towards Automated and Teleoperated Robot Cooperation in Industrial Recovery Scenarios

In this study, we present an optimization framework for efficient motion priority design between automated and teleoperated robots in an industrial recovery scenario. Although robots have recently become increasingly common in industrial sites, there are still challenges in achieving human-robot collaboration/cooperation (HRC), where human workers and robots are engaged in collaborative and cooperative tasks in a shared workspace. For example, the corresponding factory cell must be suspended for safety when an industrial robot drops an assembling part in the workspace. After that, a human worker is allowed to enter the robot workspace to address the robot recovery. This process causes non-continuous manufacturing, which leads to a productivity reduction. Recently, robotic teleoperation technology has emerged as a promising solution to enable people to perform tasks remotely and safely. This technology can be used in the recovery process in manufacturing failure scenarios. Our proposition involves the design of an appropriate priority function that aids in collision avoidance between the manufacturing and recovery robots and facilitates continuous processes with minimal production loss within an acceptable risk level. This paper presents a framework, including an HRC simulator and an optimization formulation, for finding optimal parameters of the priority function. Through quantitative and qualitative experiments, we address the proof of our novel concept and demonstrate its feasibility.

A Closed-Loop Bin Picking System for Entangled Wire Harnesses using Bimanual and Dynamic Manipulation

This paper addresses the challenge of industrial bin picking using entangled wire harnesses. Wire harnesses are essential in manufacturing but poses challenges in automation due to their complex geometries and propensity for entanglement. Our previous work tackled this issue by proposing a quasi-static pulling motion to separate the entangled wire harnesses. However, it still lacks sufficiency and generalization to various shapes and structures. In this paper, we deploy a dual-arm robot that can grasp, extract and disentangle wire harnesses from dense clutter using dynamic manipulation. The robot can swing to dynamically discard the entangled objects and regrasp to adjust the undesirable grasp pose. To improve the robustness and accuracy of the system, we leverage a closed-loop framework that uses haptic feedback to detect entanglement in real-time and flexibly adjust system parameters. Our bin picking system achieves an overall success rate of 91.2% in the real-world experiments using two different types of long wire harnesses. It demonstrates the effectiveness of our system in handling various wire harnesses for industrial bin picking.

Force Map: Learning to Predict Contact Force Distribution from Vision

When humans see a scene, they can roughly imagine the forces applied to objects based on their experience and use them to handle the objects properly. This paper considers transferring this "force-visualization" ability to robots. We hypothesize that a rough force distribution (named "force map") can be utilized for object manipulation strategies even if accurate force estimation is impossible. Based on this hypothesis, we propose a training method to predict the force map from vision. To investigate this hypothesis, we generated scenes where objects were stacked in bulk through simulation and trained a model to predict the contact force from a single image. We further applied domain randomization to make the trained model function on real images. The experimental results showed that the model trained using only synthetic images could predict approximate patterns representing the contact areas of the objects even for real images. Then, we designed a simple algorithm to plan a lifting direction using the predicted force distribution. We confirmed that using the predicted force distribution contributes to finding natural lifting directions for typical real-world scenes. Furthermore, the evaluation through simulations showed that the disturbance caused to surrounding objects was reduced by 26 % (translation displacement) and by 39 % (angular displacement) for scenes where objects were overlapping.

Learning to Dexterously Pick or Separate Tangled-Prone Objects for Industrial Bin Picking

Industrial bin picking for tangled-prone objects requires the robot to either pick up untangled objects or perform separation manipulation when the bin contains no isolated objects. The robot must be able to flexibly perform appropriate actions based on the current observation. It is challenging due to high occlusion in the clutter, elusive entanglement phenomena, and the need for skilled manipulation planning. In this paper, we propose an autonomous, effective and general approach for picking up tangled-prone objects for industrial bin picking. First, we learn PickNet - a network that maps the visual observation to pixel-wise possibilities of picking isolated objects or separating tangled objects and infers the corresponding grasp. Then, we propose two effective separation strategies: Dropping the entangled objects into a buffer bin to reduce the degree of entanglement; Pulling to separate the entangled objects in the buffer bin planned by PullNet - a network that predicts position and direction for pulling from visual input. To efficiently collect data for training PickNet and PullNet, we embrace the self-supervised learning paradigm using an algorithmic supervisor in a physics simulator. Real-world experiments show that our policy can dexterously pick up tangled-prone objects with success rates of 90%. We further demonstrate the generalization of our policy by picking a set of unseen objects. Supplementary material, code, and videos can be found at https://xinyiz0931.github.io/tangle.

Learning a Sequential Policy of Efficient Actions for Tangled-Prone Parts in Robotic Bin Picking

This paper introduces an autonomous bin picking system for cable harnesses - an extremely challenging object in bin picking task. Currently cable harnesses are unsuitable to be imported to automated production due to their length and elusive structures. Considering the task of robotic bin picking where the harnesses are heavily entangled, it is challenging for a robot to pick harnesses up one by one using conventional bin picking methods. In this paper, we present an efficient approach to overcoming the difficulties when dealing with entangled-prone parts. We develop several motion schemes for the robot to pick up a single harness avoiding any entanglement. Moreover, we proposed a learning-based bin picking policy to select both grasps and designed motion schemes in a reasonable sequence. Our method is unique due to the novelty for sufficiently solving the entanglement problem in picking cluttered cable harnesses. We demonstrate our approach on a set of real-world experiments, during which the proposed method is capable to perform the sequential bin picking task with both effectiveness and accuracy under a variety of cluttered scenarios.