Automating cloth folding is a challenging task with practical implications in various domains. Existing methods often struggle with unaligned configurations, limiting their applicability in real-world scenarios. In this research, we present FabricFlowAlignNet (FFAN), a novel approach that learns flow-based correspondences on point clouds between the current observed and goal cloth configurations. We use these learned 3D correspondences for both cloth alignment and manipulation: correspondences are used to align the observed cloth with the goal, and the flow-based correspondences are re-used as action proposals. Our experiments demonstrate that FFAN demonstrates superior performance compared to a state-of-the-art folding approach, particularly in scenarios where observed cloth is rotated or otherwise unaligned with the goal. 

Point-based Correspondence Estimation for Cloth Alignment and Manipulation

How the visual imitation learning models can generalize to novel unseen visual observations is a highly challenging problem. Such a generalization ability is very crucial for their real-world applications. Since this generalization problem has many different aspects, we focus on one case called spatial generalization, which refers to generalization to unseen setup of object~(entity) locations in a task, such as a novel setup of object locations in the robotic manipulation problem. In this case, previous works observe that the visual imitation learning models will overfit to the absolute information (e.g., coordinates) rather than the relational information between objects, which is more important for decision making. As a result, the models will perform poorly in novel object location setups. Nevertheless, so far, it remains unclear how we can solve this problem effectively. Our insight into this problem is to explicitly remove the absolute information from the features learned by imitation learning models so that the models can use robust, relational information to make decisions. To this end, we propose a novel, position-invariant regularizer called POINT for generalization. The proposed regularizer will penalize the imitation learning model when its features contain absolute, positional information of objects. Various experiments demonstrate the effectiveness of our method.

Spatial Generalization of Visual Imitation Learning with Position-Invariant Regularization

Neural motion planners (NMPs) demonstrate fast computational speed in finding path solutions but require a huge amount of expert trajectories for learning, thus adding a significant training computational load. In contrast, recent advancements have also led to a physics-informed NMP approach that directly solves the Eikonal equation for motion planning and does not require expert demonstrations for learning. However, experiments show that the physics-informed NMP approach performs poorly in complex environments and lacks scalability in high-dimensional real robot settings. To overcome these limitations, this paper presents a novel and tractable Eikonal equation formulation and introduces a new progressive learning strategy to train neural networks without expert data in complex, cluttered, high-dimensional robot motion planning scenarios. We show that our approach scales to the real robot set up in a narrow passage environment. 
The proposed method's videos and code implementations are available at https://github.com/ruiqini/P-NTFields.

Progressive Learning for Physics-informed Neural Motion Planning

Given point cloud input, the problem of 6-DoF grasp pose detection is to identify a set of hand poses in $\SE(3)$ from which an object can be successfully grasped. This important problem has many practical applications. Here we propose a novel method and neural network model that enables better grasp success rates relative to what is available in the literature. The method takes standard point cloud data as input and works well with single-view point clouds observed from arbitrary viewing directions. Videos and code are available at \url{https://haojhuang.github.io/edge_grasp_page/}.

Edge Grasp Network: A Graph-Based SE(3)-invariant Approach to Grasp Detection

This work studies the impact of morphological symmetries in learning applications in robotics. Morphological symmetries are a predominant feature in both biological and robotic systems, arising from the presence of planes/axis of symmetry in the system's morphology. This results in harmonious duplication and distribution of body parts  (e.g., humans' sagittal/left-right symmetry). Morphological symmetries become a significant learning prior as they extend to symmetries in the system's dynamics, optimal control policies, and in all proprioceptive and exteroceptive measurements, related to the system's dynamics evolution \cite{ordonez2023discrete}. Exploiting these symmetries in learning applications offers several advantageous outcomes, such as the use of data augmentation to mitigate the cost and challenges of data collection, or the use of equivariant/invariant function approximation models (e.g., neural networks) to improve sample efficiency and generalization, while reducing the number of trainable parameters. Lastly, we provide a video presentation and an open access repository reproducing our experiments and allowing for rapid prototyping in robot learning applications exploiting morphological symmetries.

Morphological symmetries in robot learning

Conventional end-to-end visual robotic manipulation learning methods often face challenges related to data inefficiency and limited generalizability.  To mitigate these challenges, recent works have proposed incorporating equivariance into their designs. This paper presents a fresh perspective on the design principles of SE(3)-equivariant methods for end-to-end visual robotic manipulation learning. Specifically, we examine the recently introduced concept of Equivariant Descriptor Fields (EDFs), focusing on four key underlying principles: generative modeling, bi-equivariance, steerable representation, and locality. These principles enable EDFs in achieving impressive data efficiency and out-of-distribution generalizability, even in the absence of prior knowledge. By comparing EDFs with other contemporary equivariant methods based on the four criteria, this paper underscores the importance of these design principles and aims to establish a guiding framework for future research on SE(3)-equivariant robotic manipulation.

Robotic Manipulation Learning with Equivariant Descriptor Fields: Generative Modeling, Bi-equivariance, Steerability, and Locality

Continual reinforcement learning aims to sequentially learn a variety of tasks, retaining the ability to perform previously encountered tasks while simultaneously developing new policies for novel tasks. However, current continual RL approaches overlook the fact that certain tasks are identical under basic group operations like rotations or translations, especially with visual inputs. They may unnecessarily learn and maintain a new policy for each similar task, leading to poor sample efficiency and weak generalization capability. To address this, we introduce a unique Continual Vision-based Reinforcement Learning method that recognizes Group Symmetries, called COVERS, cultivating a policy for each group of equivalent tasks rather than individual tasks. COVERS employs a proximal policy optimization-based RL algorithm with an equivariant feature extractor and a novel task grouping mechanism that relies on the extracted invariant features. We evaluate COVERS on sequences of table-top manipulation tasks that incorporate image observations and robot proprioceptive information in both simulations and on real robot platforms. Our results show that COVERS accurately assigns tasks to their respective groups and significantly outperforms existing methods in terms of generalization capability.

Continual Reinforcement Learning with Group Symmetries

Geometric regularity, which leverages data symmetry, has been successfully incorporated into deep learning architectures such as CNNs, RNNs, GNNs, and Transformers. While this concept has been widely applied in robotics to address the curse of dimensionality when learning from high-dimensional data, the inherent reflectional and rotational symmetry of robot structures has not been adequately explored. Drawing inspiration from cooperative multi-agent reinforcement learning, we introduce novel network structures for deep learning algorithms that explicitly capture this geometric regularity. Moreover, we investigate the relationship between the geometric prior and the concept of Parameter Sharing in multi-agent reinforcement learning. Through experiments conducted on various challenging continuous control tasks, we demonstrate the significant potential of the proposed geometric regularity in enhancing robot learning capabilities.

Geometric Regularity with Robot Intrinsic Symmetry in Reinforcement Learning

Quickly and reliably finding accurate inverse kinematics (IK) solutions remains a challenging problem for robotic manipulation. Existing numerical solvers typically only produce a single solution and rely on local search techniques to minimize highly nonconvex objective functions. Recently, learning-based approaches that approximate the entire feasible set of solutions have shown promise as a means to generate multiple fast and accurate IK results in parallel. However, existing learning-based techniques have a significant drawback: each robot of interest requires a specialized model that must be trained from scratch. To address this key shortcoming, we investigate a novel distance-geometric robot representation coupled with a graph structure that allows us to leverage the flexibility of graph neural networks (GNNs). We use this approach to train a generative graphical inverse kinematics (GGIK) solver that is able to produce a large number of diverse solutions in parallel while also generalizing well---a single learned model can be used to produce IK solutions for a variety of different robots. The graphical formulation elegantly exposes the symmetry and Euclidean equivariance of the IK problem, stemming from the spatial nature of robot manipulators. We exploit this symmetry by explicitly encoding it into the architecture of our learned model, yielding a flexible solver that is able to produce IK solution sets for multiple robots.

Euclidean Equivariant Models for Generative Graphical Inverse Kinematics

Problems involving geometric data arise in a variety of fields, including computer vision, robotics, chemistry, and physics. Such data can take numerous forms, such as points, direction vectors, planes, or transformations, but to date there is no single architecture that can be applied to such a wide variety of geometric types while respecting their symmetries. In this paper we introduce the Geometric Algebra Transformer (GATr), a general-purpose architecture for geometric data. GATr represents inputs, outputs, and hidden states in the projective geometric algebra, which offers an efficient 16-dimensional vector space representation of common geometric objects and operators acting on them. GATr is equivariant with respect to E(3), the symmetry group of 3D Euclidean space. As a transformer, GATr is scalable, expressive, and versatile. In experiments with n-body modeling and robotic planning, GATr shows strong improvements over non-geometric baselines.

RSS 2023 Workshop Symmetry Submissions