Large Language Models (LLMs) serve as the core components of AI agents used across a wide range of applications, including enterprise workflow automation, software engineering, web automation, computer use, and research. These agentic LLM inference tasks are fundamentally different from traditional chatbot-focused inference — they often have much larger context lengths to capture complex, prolonged inputs, such as an entire webpage DOM or complicated tool call trajectories. This, in turn, generates significant off-chip memory traffic for hardware at the inference stage and causes the workload to be constrained by the two memory walls, namely the bandwidth and capacity walls, preventing the compute units from achieving high utilization. In this paper, we introduce PLENA, a hardware–software co-designed system that applies three core optimization pathways. PLENA features a novel flattened systolic-array architecture (Pathway 1) and efficient compute and memory units that support an asymmetric quantization scheme (Pathway 2). It also provides native support for FlashAttention (Pathway 3). In addition, PLENA is developed with a complete software–hardware stack, including a custom ISA, a compiler, a transaction-level simulator, and an automated design-space exploration flow. Experimental results show that PLENA delivers up to 2.23× and 4.70× higher throughput than the A100 GPU and TPU v6e, respectively, under identical multiplier counts and memory configurations during LLaMA agentic inference. PLENA also achieves up to 4.04× higher energy efficiency than the A100 GPU.

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.

If a robot masters folding a kitchen towel, we would expect it to master folding a large beach towel. However, existing policy learning methods that rely on data augmentation still don’t guarantee such generalization. Our insight is to add equivariance to both the visual object representation and policy architecture. We propose EquivAct which utilizes SIM(3)-equivariant network structures that guarantee generalization across all possible object translations, 3D rotations, and scales by construction. EquivAct is trained in two phases. We first pre-train a SIM(3)-equivariant visual representation on simulated scene point clouds. Then, we learn a SIM(3)-equivariant visuomotor policy using a small amount of source task demonstrations. We show that the learned policy directly transfers to objects that substantially differ from demonstrations in scale, position, and orientation. We evaluate our method in three manipulation tasks involving deformable and articulated objects, going beyond typical rigid object manipulation tasks considered in prior work. We conduct experiments both in simulation and in reality. For real robot experiments, our method uses 20 human demonstrations of a tabletop task and transfers zero-shot to a mobile manipulation task in a much larger setup. Experiments confirm that our contrastive pre-training procedure and equivariant architecture offer significant improvements over prior work. Project website: equivact.github.io

When humans perform contact-rich manipulation tasks, customized tools are often necessary to simplify the task. For instance, we use various utensils for handling food, such as knives, forks and spoons. Similarly, robots may benefit from specialized tools that enable them to more easily complete a variety of tasks. We present an end-to-end framework to automatically learn tool morphology for contact-rich manipulation tasks by leveraging differentiable physics simulators. Previous work relied on manually constructed priors requiring detailed specification of a 3D object model, grasp pose and task description to facilitate the search or optimization process. Our approach only requires defining the objective with respect to task performance and enables learning a robust morphology through randomizing variations of the task. We make this optimization tractable by casting it as a continual learning problem. We demonstrate the effectiveness of our method for designing new tools in several scenarios, such as winding ropes, flipping a box and pushing peas onto a scoop in simulation. Additionally, experiments with real robots show that the tool shapes discovered by our method help them succeed in these scenarios.

In this paper, we present a method to manipulate unknown objects in-hand using tactile sensing without relying on a known object model. In many cases, vision-only approaches may not be feasible; for example, due to occlusion in cluttered spaces. We address this limitation by introducing a method to reorient unknown objects using tactile sensing. It incrementally builds a probabilistic estimate of the object shape and pose during task-driven manipulation. Our approach uses Bayesian optimization to balance exploration of the global object shape with efficient task completion. To demonstrate the effectiveness of our method, we apply it to a simulated Tactile-Enabled Roller Grasper, a gripper that rolls objects in hand while collecting tactile data. We evaluate our method on an insertion task with randomly generated objects and find that it reliably reorients objects while significantly reducing the exploration time.

For a robot to personalize physical assistance effectively, it must learn user preferences that can be generally reapplied to future scenarios. In this work, we investigate personalization of household cleanup with robots that can tidy up rooms by picking up objects and putting them away. A key challenge is determining the proper place to put each object, as people's preferences can vary greatly depending on personal taste or cultural background. For instance, one person may prefer storing shirts in the drawer, while another may prefer them on the shelf. We aim to build systems that can learn such preferences from just a handful of examples via prior interactions with a particular person. We show that robots can combine language-based planning and perception with the few-shot summarization capabilities of large language models (LLMs) to infer generalized user preferences that are broadly applicable to future interactions. This approach enables fast adaptation and achieves 91.2% accuracy on unseen objects in our benchmark dataset. We also demonstrate our approach on a real-world mobile manipulator called TidyBot, which successfully puts away 85.0% of objects in real-world test scenarios.

We develop a method for learning periodic tasks from visual demonstrations. The core idea is to leverage periodicity in the policy structure to model periodic aspects of the tasks. We use active learning to optimize parameters of rhythmic dynamic movement primitives (rDMPs) and propose an objective to maximize the similarity between the motion of objects manipulated by the robot and the desired motion in human video demonstrations. We consider tasks with deformable objects and granular matter whose states are challenging to represent and track: wiping surfaces with a cloth, winding cables, and stirring granular matter with a spoon. Our method does not require tracking markers or manual annotations. The initial training data consists of 10-minute videos of random unpaired interactions with objects by the robot and human. We use these for unsupervised learning of a keypoint model to get task-agnostic visual correspondences. Then, we use Bayesian optimization to optimize rDMPs from a single human video demonstration within few robot trials. We present simulation and hardware experiments to validate our approach.

Research in manipulation of deformable objects is typically conducted on a limited range of scenarios, because handling each scenario on hardware takes significant effort. Realistic simulators with support for various types of deformations and interactions have the potential to speed up experimentation with novel tasks and algorithms. However, for highly deformable objects it is challenging to align the output of a simulator with the behavior of real objects. Manual tuning is not intuitive, hence automated methods are needed. We view this alignment problem as a joint perception-inference challenge and demonstrate how to use recent neural network architectures to successfully perform simulation parameter inference from real point clouds. We analyze the performance of various architectures, comparing their data and training requirements. Furthermore, we propose to leverage differentiable point cloud sampling and differentiable simulation to significantly reduce the time to achieve the alignment. We employ an efficient way to propagate gradients from point clouds to simulated meshes and further through to the physical simulation parameters, such as mass and stiffness. Experiments with highly deformable objects show that our method can achieve comparable or better alignment with real object behavior, while reducing the time needed to achieve this by more than an order of magnitude. Videos and supplementary material are available at https://tinyurl.com/diffcloud.

General-purpose simulators can be a valuable data source for flexible learning and control approaches. However, training models or control policies in simulation and then directly applying to hardware can yield brittle control. Instead, we propose a novel way to use simulators as regularizers. Our approach regularizes a decoder of a variational autoencoder to a black-box simulation, with the latent space bound to a subset of simulator parameters. This enables successful encoder training from a small number of real-world trajectories (10 in our experiments), yielding a latent space with simulation parameter distribution that matches the real-world setting. We use a learnable mixture for the latent prior/posterior, which implies a highly flexible class of densities for the posterior fit. Our approach is scalable and does not require restrictive distributional assumptions. We demonstrate ability to recover matching parameter distributions on a range of benchmarks, challenging custom simulation environments and several real-world scenarios. Our experiments using ABB YuMi robot hardware show ability to help reinforcement learning approaches overcome cases of severe sim-to-real mismatch.

Learning for control can acquire controllers for novel robotic tasks, paving the path for autonomous agents. Such controllers can be expert-designed policies, which typically require tuning of parameters for each task scenario. In this context, Bayesian optimization (BO) has emerged as a promising approach for automatically tuning controllers. However, sample-efficiency can still be an issue for high-dimensional policies on hardware. Here, we develop an approach that utilizes simulation to learn structured feature transforms that map the original parameter space into a domain-informed space. During BO, similarity between controllers is now calculated in this transformed space. Experiments on the ATRIAS robot hardware and simulation show that our approach succeeds at sample-efficiently learning controllers for multiple robots. Another question arises: What if the simulation significantly differs from hardware? To answer this, we create increasingly approximate simulators and study the effect of increasing simulation-hardware mismatch on the performance of Bayesian optimization. We also compare our approach to other approaches from literature, and find it to be more reliable, especially in cases of high mismatch. Our experiments show that our approach succeeds across different controller types, bipedal robot models and simulator fidelity levels, making it applicable to a wide range of bipedal locomotion problems.

Robotics controllers often consist of expert-designed heuristics, which can be hard to tune in higher dimensions. Simulation can aid in optimizing these controllers if parameters learned in simulation transfer to hardware. Unfortunately, this is often not the case in legged locomotion, necessitating learning directly on hardware. This motivates using data-efficient learning techniques like Bayesian Optimization (BO) to minimize collecting expensive data samples. BO is a black-box data-efficient optimization scheme, though its performance typically degrades in higher dimensions. We aim to overcome this problem by incorporating domain knowledge, with a focus on bipedal locomotion. In our previous work, we proposed a feature transformation that projected a 16-dimensional locomotion controller to a 1-dimensional space using knowledge of human walking. When optimizing a human-inspired neuromuscular controller in simulation, this feature transformation enhanced sample efficiency of BO over traditional BO with a Squared Exponential kernel. In this paper, we present a generalized feature transform applicable to non-humanoid robot morphologies and evaluate it on the ATRIAS bipedal robot, in both simulation and hardware. We present three different walking controllers and two are evaluated on the real robot. Our results show that this feature transform captures important aspects of walking and accelerates learning on hardware and simulation, as compared to traditional BO.

We seek to automatically identify which items to include in a set of curriculum, and how to adaptively select these items, in order to maximize student performance on some specified set of learning objectives. Our experimental results with a histogram tutoring system suggest that Bayesian Optimization can quickly (with only a small amount of student data) find good parameters, and may help instructors identify misalignment between their course, and their desired learning objectives.