Scaffolding Dexterous Manipulation with Vision-Language Models

Dexterous robotic hands are essential for performing complex manipulation tasks, yet remain difficult to train due to the challenges of demonstration collection and high-dimensional control. While reinforcement learning (RL) can alleviate the data bottleneck by generating experience in simulation, it typically relies on carefully designed, task-specific reward functions, which hinder scalability and generalization. Thus, contemporary works in dexterous manipulation have often bootstrapped from reference trajectories. These trajectories specify target hand poses that guide the exploration of RL policies and object poses that enable dense, task-agnostic rewards. However, sourcing suitable trajectories - particularly for dexterous hands - remains a significant challenge. Yet, the precise details in explicit reference trajectories are often unnecessary, as RL ultimately refines the motion. Our key insight is that modern vision-language models (VLMs) already encode the commonsense spatial and semantic knowledge needed to specify tasks and guide exploration effectively. Given a task description (e.g., "open the cabinet") and a visual scene, our method uses an off-the-shelf VLM to first identify task-relevant keypoints (e.g., handles, buttons) and then synthesize 3D trajectories for hand motion and object motion. Subsequently, we train a low-level residual RL policy in simulation to track these coarse trajectories or "scaffolds" with high fidelity. Across a number of simulated tasks involving articulated objects and semantic understanding, we demonstrate that our method is able to learn robust dexterous manipulation policies. Moreover, we showcase that our method transfers to real-world robotic hands without any human demonstrations or handcrafted rewards.

CUPID: Curating Data your Robot Loves with Influence Functions

In robot imitation learning, policy performance is tightly coupled with the quality and composition of the demonstration data. Yet, developing a precise understanding of how individual demonstrations contribute to downstream outcomes - such as closed-loop task success or failure - remains a persistent challenge. We propose CUPID, a robot data curation method based on a novel influence function-theoretic formulation for imitation learning policies. Given a set of evaluation rollouts, CUPID estimates the influence of each training demonstration on the policy's expected return. This enables ranking and selection of demonstrations according to their impact on the policy's closed-loop performance. We use CUPID to curate data by 1) filtering out training demonstrations that harm policy performance and 2) subselecting newly collected trajectories that will most improve the policy. Extensive simulated and hardware experiments show that our approach consistently identifies which data drives test-time performance. For example, training with less than 33% of curated data can yield state-of-the-art diffusion policies on the simulated RoboMimic benchmark, with similar gains observed in hardware. Furthermore, hardware experiments show that our method can identify robust strategies under distribution shift, isolate spurious correlations, and even enhance the post-training of generalist robot policies. Additional materials are made available at: https://cupid-curation.github.io.

Vision in Action: Learning Active Perception from Human Demonstrations

We present Vision in Action (ViA), an active perception system for bimanual robot manipulation. ViA learns task-relevant active perceptual strategies (e.g., searching, tracking, and focusing) directly from human demonstrations. On the hardware side, ViA employs a simple yet effective 6-DoF robotic neck to enable flexible, human-like head movements. To capture human active perception strategies, we design a VR-based teleoperation interface that creates a shared observation space between the robot and the human operator. To mitigate VR motion sickness caused by latency in the robot's physical movements, the interface uses an intermediate 3D scene representation, enabling real-time view rendering on the operator side while asynchronously updating the scene with the robot's latest observations. Together, these design elements enable the learning of robust visuomotor policies for three complex, multi-stage bimanual manipulation tasks involving visual occlusions, significantly outperforming baseline systems.

Mobi-$π$: Mobilizing Your Robot Learning Policy

Learned visuomotor policies are capable of performing increasingly complex manipulation tasks. However, most of these policies are trained on data collected from limited robot positions and camera viewpoints. This leads to poor generalization to novel robot positions, which limits the use of these policies on mobile platforms, especially for precise tasks like pressing buttons or turning faucets. In this work, we formulate the policy mobilization problem: find a mobile robot base pose in a novel environment that is in distribution with respect to a manipulation policy trained on a limited set of camera viewpoints. Compared to retraining the policy itself to be more robust to unseen robot base pose initializations, policy mobilization decouples navigation from manipulation and thus does not require additional demonstrations. Crucially, this problem formulation complements existing efforts to improve manipulation policy robustness to novel viewpoints and remains compatible with them. To study policy mobilization, we introduce the Mobi-$\pi$ framework, which includes: (1) metrics that quantify the difficulty of mobilizing a given policy, (2) a suite of simulated mobile manipulation tasks based on RoboCasa to evaluate policy mobilization, (3) visualization tools for analysis, and (4) several baseline methods. We also propose a novel approach that bridges navigation and manipulation by optimizing the robot's base pose to align with an in-distribution base pose for a learned policy. Our approach utilizes 3D Gaussian Splatting for novel view synthesis, a score function to evaluate pose suitability, and sampling-based optimization to identify optimal robot poses. We show that our approach outperforms baselines in both simulation and real-world environments, demonstrating its effectiveness for policy mobilization.

Causal-PIK: Causality-based Physical Reasoning with a Physics-Informed Kernel

Tasks that involve complex interactions between objects with unknown dynamics make planning before execution difficult. These tasks require agents to iteratively improve their actions after actively exploring causes and effects in the environment. For these type of tasks, we propose Causal-PIK, a method that leverages Bayesian optimization to reason about causal interactions via a Physics-Informed Kernel to help guide efficient search for the best next action. Experimental results on Virtual Tools and PHYRE physical reasoning benchmarks show that Causal-PIK outperforms state-of-the-art results, requiring fewer actions to reach the goal. We also compare Causal-PIK to human studies, including results from a new user study we conducted on the PHYRE benchmark. We find that Causal-PIK remains competitive on tasks that are very challenging, even for human problem-solvers.

SLAG: Scalable Language-Augmented Gaussian Splatting

Language-augmented scene representations hold great promise for large-scale robotics applications such as search-and-rescue, smart cities, and mining. Many of these scenarios are time-sensitive, requiring rapid scene encoding while also being data-intensive, necessitating scalable solutions. Deploying these representations on robots with limited computational resources further adds to the challenge. To address this, we introduce SLAG, a multi-GPU framework for language-augmented Gaussian splatting that enhances the speed and scalability of embedding large scenes. Our method integrates 2D visual-language model features into 3D scenes using SAM and CLIP. Unlike prior approaches, SLAG eliminates the need for a loss function to compute per-Gaussian language embeddings. Instead, it derives embeddings from 3D Gaussian scene parameters via a normalized weighted average, enabling highly parallelized scene encoding. Additionally, we introduce a vector database for efficient embedding storage and retrieval. Our experiments show that SLAG achieves an 18 times speedup in embedding computation on a 16-GPU setup compared to OpenGaussian, while preserving embedding quality on the ScanNet and LERF datasets. For more details, visit our project website: https://slag-project.github.io/.

Deformable Cargo Transport in Microgravity with Astrobee

We present pyastrobee: a simulation environment and control stack for Astrobee in Python, with an emphasis on cargo manipulation and transport tasks. We also demonstrate preliminary success from a sampling-based MPC controller, using reduced-order models of NASA's cargo transfer bag (CTB) to control a high-order deformable finite element model. Our code is open-source, fully documented, and available at https://danielpmorton.github.io/pyastrobee

Crossing the Human-Robot Embodiment Gap with Sim-to-Real RL using One Human Demonstration

Teaching robots dexterous manipulation skills often requires collecting hundreds of demonstrations using wearables or teleoperation, a process that is challenging to scale. Videos of human-object interactions are easier to collect and scale, but leveraging them directly for robot learning is difficult due to the lack of explicit action labels from videos and morphological differences between robot and human hands. We propose Human2Sim2Robot, a novel real-to-sim-to-real framework for training dexterous manipulation policies using only one RGB-D video of a human demonstrating a task. Our method utilizes reinforcement learning (RL) in simulation to cross the human-robot embodiment gap without relying on wearables, teleoperation, or large-scale data collection typically necessary for imitation learning methods. From the demonstration, we extract two task-specific components: (1) the object pose trajectory to define an object-centric, embodiment-agnostic reward function, and (2) the pre-manipulation hand pose to initialize and guide exploration during RL training. We found that these two components are highly effective for learning the desired task, eliminating the need for task-specific reward shaping and tuning. We demonstrate that Human2Sim2Robot outperforms object-aware open-loop trajectory replay by 55% and imitation learning with data augmentation by 68% across grasping, non-prehensile manipulation, and multi-step tasks. Project Site: https://human2sim2robot.github.io

Force-Aware Autonomous Robotic Surgery

This work demonstrates the benefits of using tool-tissue interaction forces in the design of autonomous systems in robot-assisted surgery (RAS). Autonomous systems in surgery must manipulate tissues of different stiffness levels and hence should apply different levels of forces accordingly. We hypothesize that this ability is enabled by using force measurements as input to policies learned from human demonstrations. To test this hypothesis, we use Action-Chunking Transformers (ACT) to train two policies through imitation learning for automated tissue retraction with the da Vinci Research Kit (dVRK). To quantify the effects of using tool-tissue interaction force data, we trained a "no force policy" that uses the vision and robot kinematic data, and compared it to a "force policy" that uses force, vision and robot kinematic data. When tested on a previously seen tissue sample, the force policy is 3 times more successful in autonomously performing the task compared with the no force policy. In addition, the force policy is more gentle with the tissue compared with the no force policy, exerting on average 62% less force on the tissue. When tested on a previously unseen tissue sample, the force policy is 3.5 times more successful in autonomously performing the task, exerting an order of magnitude less forces on the tissue, compared with the no force policy. These results open the door to design force-aware autonomous systems that can meet the surgical guidelines for tissue handling, especially using the newly released RAS systems with force feedback capabilities such as the da Vinci 5.

DexForce: Extracting Force-informed Actions from Kinesthetic Demonstrations for Dexterous Manipulation

Imitation learning requires high-quality demonstrations consisting of sequences of state-action pairs. For contact-rich dexterous manipulation tasks that require fine-grained dexterity, the actions in these state-action pairs must produce the right forces. Current widely-used methods for collecting dexterous manipulation demonstrations are difficult to use for demonstrating contact-rich tasks due to unintuitive human-to-robot motion retargeting and the lack of direct haptic feedback. Motivated by this, we propose DexForce, a method for collecting demonstrations of contact-rich dexterous manipulation. DexForce leverages contact forces, measured during kinesthetic demonstrations, to compute force-informed actions for policy learning. We use DexForce to collect demonstrations for six tasks and show that policies trained on our force-informed actions achieve an average success rate of 76% across all tasks. In contrast, policies trained directly on actions that do not account for contact forces have near-zero success rates. We also conduct a study ablating the inclusion of force data in policy observations. We find that while using force data never hurts policy performance, it helps the most for tasks that require an advanced level of precision and coordination, like opening an AirPods case and unscrewing a nut.