Preference-based reinforcement learning (PbRL) is a suitable approach for style adaptation of pre-trained robotic behavior: adapting the robot's policy to follow human user preferences while still being able to perform the original task. However, collecting preferences for the adaptation process in robotics is often challenging and time-consuming. In this work we explore the adaptation of pre-trained robots in the low-preference-data regime. We show that, in this regime, recent adaptation approaches suffer from catastrophic reward forgetting (CRF), where the updated reward model overfits to the new preferences, leading the agent to become unable to perform the original task. To mitigate CRF, we propose to enhance the original reward model with a small number of parameters (low-rank matrices) responsible for modeling the preference adaptation. Our evaluation shows that our method can efficiently and effectively adjust robotic behavior to human preferences across simulation benchmark tasks and multiple real-world robotic tasks. We provide videos of our results and source code at https://sites.google.com/view/preflora/

We address prehensile pushing, the problem of manipulating a grasped object by pushing against the environment. Our solution is an efficient nonlinear trajectory optimization problem relaxed from an exact mixed integer non-linear trajectory optimization formulation. The critical insight is recasting the external pushers (environment) as a discrete probability distribution instead of binary variables and minimizing the entropy of the distribution. The probabilistic reformulation allows all pushers to be used simultaneously, but at the optimum, the probability mass concentrates onto one due to the entropy minimization. We numerically compare our method against a state-of-the-art sampling-based baseline on a prehensile pushing task. The results demonstrate that our method finds trajectories 8 times faster and at a 20 times lower cost than the baseline. Finally, we demonstrate that a simulated and real Frank Panda robot can successfully manipulate different objects following the trajectories proposed by our method.

We introduce Cloth-Splatting, a method for estimating 3D states of cloth from RGB images through a prediction-update framework. Cloth-Splatting leverages an action-conditioned dynamics model for predicting future states and uses 3D Gaussian Splatting to update the predicted states. Our key insight is that coupling a 3D mesh-based representation with Gaussian Splatting allows us to define a differentiable map between the cloth's state space and the image space. This enables the use of gradient-based optimization techniques to refine inaccurate state estimates using only RGB supervision. Our experiments demonstrate that Cloth-Splatting not only improves state estimation accuracy over current baselines but also reduces convergence time by ∼85 %.

We introduce DexDiffuser, a novel dexterous grasping method that generates, evaluates, and refines grasps on partial object point clouds. DexDiffuser includes the conditional diffusion-based grasp sampler DexSampler and the dexterous grasp evaluator DexEvaluator. DexSampler generates high-quality grasps conditioned on object point clouds by iterative denoising of randomly sampled grasps. We also introduce two grasp refinement strategies: Evaluator-Guided Diffusion and Evaluator-based Sampling Refinement. The experiment results demonstrate that DexDiffuser consistently outperforms the state-of-the-art multi-finger grasp generation method FFHNet with an, on average, 9.12% and 19.44% higher grasp success rate in simulation and real robot experiments, respectively.

Most state-of-the-art data-driven grasp sampling methods propose stable and collision-free grasps uniformly on the target object. For bin-picking, executing any of those reachable grasps is sufficient. However, for completing specific tasks, such as squeezing out liquid from a bottle, we want the grasp to be on a specific part of the object's body while avoiding other locations, such as the cap. This work presents a generative grasp sampling network, VCGS, capable of constrained 6-Degrees of Freedom (DoF) grasp sampling. In addition, we also curate a new dataset designed to train and evaluate methods for constrained grasping. The new dataset, called CONG, consists of over 14 million training samples of synthetically rendered point clouds and grasps at random target areas on 2889 objects. VCGS is benchmarked against GraspNet, a state-of-the-art unconstrained grasp sampler, in simulation and on a real robot. The results demonstrate that VCGS achieves a 10-15% higher grasp success rate than the baseline while being 2-3 times as sample efficient. Supplementary material is available on our project website.

Endowing robots with tactile capabilities opens up new possibilities for their interaction with the environment, including the ability to handle fragile and/or soft objects. In this work, we equip the robot gripper with low-cost vision-based tactile sensors and propose a manipulation algorithm that adapts to both rigid and soft objects without requiring any knowledge of their properties. The algorithm relies on a touch and slip detection method, which considers the variation in the tactile images with respect to reference ones. We validate the approach on seven different objects, with different properties in terms of rigidity and fragility, to perform unplugging and lifting tasks. Furthermore, to enhance applicability, we combine the manipulation algorithm with a grasp sampler for the task of finding and picking a grape from a bunch without damaging it.

This work addresses the problem of learning approach-constrained data-driven grasp samplers. To this end, we propose GoNet: a generative grasp sampler that can constrain the grasp approach direction to a subset of SO(3). The key insight is to discretize SO(3) into a predefined number of bins and train GoNet to generate grasps whose approach directions are within those bins. At run-time, the bin aligning with the second largest principal component of the observed point cloud is selected. GoNet is benchmarked against GraspNet, a state-of-the-art unconstrained grasp sampler, in an unconfined grasping experiment in simulation and on an unconfined and confined grasping experiment in the real world. The results demonstrate that GoNet achieves higher success-over-coverage in simulation and a 12%-18% higher success rate in real-world table-picking and shelf-picking tasks than the baseline.

Evaluation of grasps on deformable 3D objects is a little-studied problem, even if the applicability of rigid object grasp quality measures for deformable ones is an open question. A central issue with most quality measures is their dependence on contact points, which for deformable objects depend on the deformations. This paper proposes a grasp quality measure for deformable objects that uses information about object deformation to calculate the grasp quality. Grasps are evaluated by simulating the deformations during grasping and predicting the contacts between the gripper and the grasped object. The contact information is then used as input for a new grasp quality metric to quantify the grasp quality. The approach is benchmarked against two classical rigid-body quality metrics on over 600 grasps in the Isaac gym simulation and over 50 real-world grasps. Experimental results show an average improvement of 18% in the grasp success rate for deformable objects compared to the classical rigid-body quality metrics. Furthermore, the proposed approach is approximately fifteen times faster to calculate than the shake task, which, to date, is one of the most reliable approaches to quantify a grasp on a deformable object.