Abstract--In this paper, we present a novel method for achieving dexterous manipulation of complex objects, while simultaneously securing the object without the use of passive support surfaces. We posit that a key difficulty for training such policies in a Reinforcement Learning framework is the difficulty of exploring the problem state space, as the accessible regions of this space form a complex structure along manifolds of a high-dimensional space. To address this challenge, we use two versions of the non-holonomic Rapidly-Exploring Random Trees algorithm; one version is more general, but requires explicit use of the environment's transition function, while the second version uses manipulation-specific kinematic constraints to attain better sample efficiency. In both cases, we use states found via sampling-based exploration to generate reset distributions that enable training control policies under full dynamic constraints via model-free Reinforcement Learning. We show that these policies are effective at manipulation problems of higher difficulty than previously shown, and also transfer effectively to real robots. Figure 1: Our method enables finger-gaiting manipulation of concave A number of example videos can also be found on the project or elongated objects which require complex gaits. Reinforcement Learning (RL) of robot sensorimotor control policies has seen great advances in recent years, demonstrated and highly effective family of Sampling-Based Planning (SBP) for a wide range of motor tasks.

Shape Memory Alloys (SMAs) are a class of smart materials that exhibit a macroscopic contraction of up to 5% when heated via an electric current. This effect can be exploited for the development of novel unconventional actuators. Despite having many features such as compactness, lightweight, and high energy density, commercial SMA wires are characterized by a highly nonlinear behavior, which manifests itself as a load-, temperature-, and rate-dependent hysteresis exhibiting a complex shape and minor loops. Accurate modeling and compensation of such hysteresis are fundamental for the development of high-performance SMA applications. In this work, we propose a new dynamical model to describe the complex hysteresis of polycrystalline SMA wires. The approach is based on a reformulation of the Muller-Achenbach-Seelecke model for uniaxial SMA wires within a hybrid dynamical framework. In this way, we can significantly reduce the numerical complexity and computation time without losing accuracy and physical interpretability. After describing the model, an extensive experimental validation campaign is carried out on a 75 {\mu}m diameter SMA wire specimen. The new hybrid model will pave the development of hybrid controllers and observers for SMA actuators.

Abstract-- We present a novel controller for fixed-wing UAVs that enables autonomous soaring in an orographic wind field, extending flight endurance. Our method identifies soaring regions and addresses position control challenges by introducing a target gradient line (TGL) on which the UAV achieves an equilibrium soaring position, where sink rate and updraft are balanced. We also demonstrate a single degree of control freedom in a soaring position through manipulation of the TGL. I. INTRODUCTION UAVs have benefited from advancements in battery technology and miniaturization of avionics, which resulted in an increase in their endurance and range. However, the full potential of UAV applications remains limited by reduced flight time.

In this letter, we address blockage detection and precoder design for multiple-input multiple-output (MIMO) links, without communication overhead required. Blockage detection is achieved by classifying light detection and ranging (LIDAR) data through a physics-based graph neural network (GNN). For precoder design, a preliminary channel estimate is obtained by running ray tracing on a 3D surface obtained from LIDAR data. This estimate is successively refined and the precoder is designed accordingly. Numerical simulations show that blockage detection is successful with 95% accuracy. Our digital precoding achieves 90% of the capacity and analog precoding outperforms previous works exploiting LIDAR for precoder design.

The properties of lithium metal are key parameters in the design of lithium ion and lithium metal batteries. They are difficult to probe experimentally due to the high reactivity and low melting point of lithium as well as the microscopic scales at which lithium exists in batteries where it is found to have enhanced strength, with implications for dendrite suppression strategies. Computationally, there is a lack of empirical potentials that are consistently quantitatively accurate across all properties and ab-initio calculations are too costly. In this work, we train Machine Learning Interaction Potentials (MLIPs) on Density Functional Theory (DFT) data to state-of-the-art accuracy in reproducing experimental and ab-initio results across a wide range of simulations at large length and time scales. We accurately predict thermodynamic properties, phonon spectra, temperature dependence of elastic constants and various surface properties inaccessible using DFT. We establish that there exists a Bell-Evans-Polanyi relation correlating the self-adsorption energy and the minimum surface diffusion barrier for high Miller index facets.

In recent years, large language models (LMs) have achieved remarkable progress across various natural language processing tasks. As pre-training and fine-tuning are costly and might negatively impact model performance, it is desired to efficiently adapt an existing model to different conditions such as styles, sentiments or narratives, when facing different audiences or scenarios. However, efficient adaptation of a language model to diverse conditions remains an open challenge. This work is inspired by the observation that text conditions are often associated with selection of certain words in a context. Therefore we introduce LM-Switch, a theoretically grounded, lightweight and simple method for generative language model conditioning. We begin by investigating the effect of conditions in Hidden Markov Models (HMMs), and establish a theoretical connection with language model. Our finding suggests that condition shifts in HMMs are associated with linear transformations in word embeddings. LM-Switch is then designed to deploy a learnable linear factor in the word embedding space for language model conditioning. We show that LM-Switch can model diverse tasks, and achieves comparable or better performance compared with state-of-the-art baselines in LM detoxification and generation control, despite requiring no more than 1% of parameters compared with baselines and little extra time overhead compared with base LMs. It is also able to learn from as few as a few sentences or one document. Moreover, a learned LM-Switch can be transferred to other LMs of different sizes, achieving a detoxification performance similar to the best baseline. We will make our code available to the research community following publication.

In this paper, a modified version of conservative Physics-informed Neural Networks (cPINN for short) is provided to construct the weak solutions of Riemann problem for the hyperbolic scalar conservation laws in non-conservative form. To demonstrate the results, we use the model of generalized Buckley-Leverett equation (GBL equation for short) with discontinuous porosity in porous media. By inventing a new unknown, the GBL equation is transformed into a two-by-two resonant hyperbolic conservation laws in conservative form. The modified method of cPINN is invented to overcome the difficulties due to the discontinuity of the porosity and the appearance of the critical states (near vacuum) in the Riemann data. We experiment with our idea by using a deep learning algorithm to solve the GBL equation in both conservative and non-conservative forms, as well as the cases of critical and non-critical states. This method provides a combination of two different neural networks and corresponding loss functions, one is for the two-by-two resonant hyperbolic system, and the other is for the scalar conservation law with a discontinuous perturbation term in the non-convex flux. The technique of re-scaling to the unknowns is adopted to avoid the oscillation of the Riemann solutions in the cases of critical Riemann data. The solutions constructed by the modified cPINN match the exact solutions constructed by the theoretical analysis for hyperbolic conservation laws. In addition, the solutions are identical in both conservative and non-conservative cases. Finally, we compare the performance of the modified cPINN with numerical method called WENO5. Whereas WENO5 struggles with the highly oscillation of approximate solutions for the Riemann problems of GBL equation in non-conservative form, cPINN works admirably.

Peking University Abstract: The backpropagation algorithm, despite its widespread use in neural network learning, may not accurately emulate the human cortex's learning process. However, the original FFA paper and related works on the Forward-Forward Algorithm only mentioned very limited types of neural network mechanisms and may limit its application and effectiveness. In response to these challenges, we propose an integrated method that combines the strengths of both FFA and shallow backpropagation, yielding a biologically plausible neural network training algorithm which can also be applied to various network structures. We applied this integrated approach to the classification of the Modified National Institute of Standards and Technology (MNIST) database, where it outperformed FFA and demonstrated superior resilience to noise compared to backpropagation. We show that training neural networks with the Integrated Forward-Forward Algorithm has the potential of generating neural networks with advantageous features like robustness. 1. Introduction For the past decade, Deep learning has made significant progress on countless tasks and problems, with backpropagation as a key contributing factor to train the deep models developed. However, in aspects of biological plausibility, although the model of deep neural networks for artificial intelligence is initially inspired by biological neurons' structures, the training process of these deep neural networks using backpropagation throughout the whole neural network does not seem biologically plausible. Researchers have argued that the long range, or global, backpropagation of gradient information has little evidence in neuron science and violates many principles of the cortex learning found in neural science, like locality and onlinety. The original Forward-Forward Algorithm is one of the approaches in biological plausible learning as an alternative of backpropagation to train neural networks initially proposed by Hinton(2022) [1].

Large swarms of extremely simple robots (i.e., capable just of basic motion activities, like propelling forward or self-rotating) are widely applied to study collective task performance based on self-organization or local algorithms instead of sophisticated programming and global swarm coordination. Moreover, they represent a versatile yet affordable platform for experimental studies in physics, particularly in active matter - non-equilibrium assemblies of particles converting their energy to a directed motion. However, a large set of robotics platforms is being used in different studies, while the universal design is still lacking. Despite such platforms possess advantages in certain application scenarios, their large number sufficiently limits further development of results in the field, as advancing some study requires to buy or manually produce the corresponding robots. To address this issue, we develop an open-source Swarmodroid 1.0 platform based on bristle-bots with reconfigurable 3D-printed bodies, external control of motion velocity, and basic capabilities of velocity profile programming. In addition, we introduce AMPy software package in Python featuring OpenCV-based extraction of robotic swarm kinematics accompanied by the evaluation of key physical quantities describing the collective dynamics. We perform a detailed analysis of individual Swarmodroids' motion characteristics and address their use cases with two examples: a cargo transport performed by self-rotating robots and a velocity-dependent jam formation in a bottleneck by self-propelling robots. Finally, we provide a comparison of existing centimeter-scale robotic platforms, a review of key quantities describing collective dynamics of many-particle systems, and a comprehensive outlook considering potential applications as well as further directions for fundamental studies and Swarmodroid 1.0 platform development.

Classical algorithms are often not effective for solving nonconvex optimization problems where local minima are separated by high barriers. In this paper, we explore possible quantum speedups for nonconvex optimization by leveraging the global effect of quantum tunneling. Specifically, we introduce a quantum algorithm termed the quantum tunneling walk (QTW) and apply it to nonconvex problems where local minima are approximately global minima. We show that QTW achieves quantum speedup over classical stochastic gradient descents (SGD) when the barriers between different local minima are high but thin and the minima are flat. Based on this observation, we construct a specific double-well landscape, where classical algorithms cannot efficiently hit one target well knowing the other well but QTW can when given proper initial states near the known well. Finally, we corroborate our findings with numerical experiments.