This paper presents a novel approach to enhance the Binary-Addition-Tree algorithm (BAT) by integrating incremental learning techniques. BAT, known for its simplicity in development, implementation, and application, is a powerful implicit enumeration method for solving network reliability and optimization problems. However, it traditionally struggles with dynamic and large-scale networks due to its static nature. By introducing incremental learning, we enable the BAT to adapt and improve its performance iteratively as it encounters new data or network changes. This integration allows for more efficient computation, reduced redundancy without searching minimal paths and cuts, and improves overall performance in dynamic environments. Experimental results demonstrate the effectiveness of the proposed method, showing significant improvements in both computational efficiency and solution quality compared to the traditional BAT and indirect algorithms, such as MP-based algorithms and MC-based algorithms.

-- Learning-based algorithms have demonstrated impressive performance in agile locomotion of legged robots. However, learned policies are often complex and opaque due to the black-box nature of learning algorithms, which hinders predictability and precludes guarantees on performance or safety. In this work, we develop a novel safe navigation framework that combines Koopman operators and model-predictive control (MPC) frameworks. Our method adopts Koopman operator theory to learn the linear evolution of dynamics of the underlying locomotion policy, which can be effectively learned with Dynamic Mode Decomposition (DMD). Given that our learned model is linear, we can readily leverage the standard MPC algorithm. Our framework is easy to implement with less prior knowledge because it does not require access to the underlying dynamical systems or control-theoretic techniques. We demonstrate that the learned linear dynamics can better predict the trajectories of legged robots than baselines. In addition, we showcase that the proposed navigation framework can achieve better safety with less collisions in challenging and dense environments with narrow passages. I. INTRODUCTION Recent advances in reinforcement learning have led to significant improvements in robust and agile quadrupedal locomotion [1]-[6].

Linear control models have gained extensive application in vehicle control due to their simplicity, ease of use, and support for stability analysis. However, these models lack adaptability to the changing environment and multi-objective settings. Reinforcement learning (RL) models, on the other hand, offer adaptability but suffer from a lack of interpretability and generalization capabilities. This paper aims to develop a family of RL-based controllers enhanced by physics-informed policies, leveraging the advantages of both physics-based models (data-efficient and interpretable) and RL methods (flexible to multiple objectives and fast computing). We propose the Physics-Enhanced Residual Policy Learning (PERPL) framework, where the physics component provides model interpretability and stability. The learning-based Residual Policy adjusts the physics-based policy to adapt to the changing environment, thereby refining the decisions of the physics model. We apply our proposed model to decentralized control to mixed traffic platoon of Connected and Automated Vehicles (CAVs) and Human-driven Vehicles (HVs) using a constant time gap (CTG) strategy for cruising and incorporating actuator and communication delays. Experimental results demonstrate that our method achieves smaller headway errors and better oscillation dampening than linear models and RL alone in scenarios with artificially extreme conditions and real preceding vehicle trajectories. At the macroscopic level, overall traffic oscillations are also reduced as the penetration rate of CAVs employing the PERPL scheme increases.

Due to high dimensionality and non-convexity, real-time optimal control using full-order dynamics models for legged robots is challenging. Therefore, Nonlinear Model Predictive Control (NMPC) approaches are often limited to reduced-order models. Sampling-based MPC has shown potential in nonconvex even discontinuous problems, but often yields suboptimal solutions with high variance, which limits its applications in high-dimensional locomotion. This work introduces DIAL-MPC (Diffusion-Inspired Annealing for Legged MPC), a sampling-based MPC framework with a novel diffusion-style annealing process. Such an annealing process is supported by the theoretical landscape analysis of Model Predictive Path Integral Control (MPPI) and the connection between MPPI and single-step diffusion. Algorithmically, DIAL-MPC iteratively refines solutions online and achieves both global coverage and local convergence. In quadrupedal torque-level control tasks, DIAL-MPC reduces the tracking error of standard MPPI by $13.4$ times and outperforms reinforcement learning (RL) policies by $50\%$ in challenging climbing tasks without any training. In particular, DIAL-MPC enables precise real-world quadrupedal jumping with payload. To the best of our knowledge, DIAL-MPC is the first training-free method that optimizes over full-order quadruped dynamics in real-time.

Bipedal locomotion control is essential for humanoid robots to navigate complex, human-centric environments. While optimization-based control designs are popular for integrating sophisticated models of humanoid robots, they often require labor-intensive manual tuning. In this work, we address the challenges of parameter selection in bipedal locomotion control using DiffTune, a model-based autotuning method that leverages differential programming for efficient parameter learning. A major difficulty lies in balancing model fidelity with differentiability. We address this difficulty using a low-fidelity model for differentiability, enhanced by a Ground Reaction Force-and-Moment Network (GRFM-Net) to capture discrepancies between MPC commands and actual control effects. We validate the parameters learned by DiffTune with GRFM-Net in hardware experiments, which demonstrates the parameters' optimality in a multi-objective setting compared with baseline parameters, reducing the total loss by up to 40.5$\%$ compared with the expert-tuned parameters. The results confirm the GRFM-Net's effectiveness in mitigating the sim-to-real gap, improving the transferability of simulation-learned parameters to real hardware.

This paper considers the distributed online bandit optimization problem with nonconvex loss functions over a time-varying digraph. This problem can be viewed as a repeated game between a group of online players and an adversary. At each round, each player selects a decision from the constraint set, and then the adversary assigns an arbitrary, possibly nonconvex, loss function to this player. Only the loss value at the current round, rather than the entire loss function or any other information (e.g. gradient), is privately revealed to the player. Players aim to minimize a sequence of global loss functions, which are the sum of local losses. We observe that traditional multi-point bandit algorithms are unsuitable for online optimization, where the data for the loss function are not all a priori, while the one-point bandit algorithms suffer from poor regret guarantees. To address these issues, we propose a novel one-point residual feedback distributed online algorithm. This algorithm estimates the gradient using residuals from two points, effectively reducing the regret bound while maintaining $\mathcal{O}(1)$ sampling complexity per iteration. We employ a rigorous metric, dynamic regret, to evaluate the algorithm's performance. By appropriately selecting the step size and smoothing parameters, we demonstrate that the expected dynamic regret of our algorithm is comparable to existing algorithms that use two-point feedback, provided the deviation in the objective function sequence and the path length of the minimization grows sublinearly. Finally, we validate the effectiveness of the proposed algorithm through numerical simulations.

Classical manipulator motion planners work across different robot embodiments. However they plan on a pre-specified static environment representation, and are not scalable to unseen dynamic environments. Neural Motion Planners (NMPs) are an appealing alternative to conventional planners as they incorporate different environmental constraints to learn motion policies directly from raw sensor observations. Contemporary state-of-the-art NMPs can successfully plan across different environments. However none of the existing NMPs generalize across robot embodiments. In this paper we propose Cross-Embodiment Motion Policy (XMoP), a neural policy for learning to plan over a distribution of manipulators. XMoP implicitly learns to satisfy kinematic constraints for a distribution of robots and $\textit{zero-shot}$ transfers the planning behavior to unseen robotic manipulators within this distribution. We achieve this generalization by formulating a whole-body control policy that is trained on planning demonstrations from over three million procedurally sampled robotic manipulators in different simulated environments. Despite being completely trained on synthetic embodiments and environments, our policy exhibits strong sim-to-real generalization across manipulators with different kinematic variations and degrees of freedom with a single set of frozen policy parameters. We evaluate XMoP on $7$ commercial manipulators and show successful cross-embodiment motion planning, achieving an average $70\%$ success rate on baseline benchmarks. Furthermore, we demonstrate our policy sim-to-real on two unseen manipulators solving novel planning problems across three real-world domains even with dynamic obstacles.

We use machine learning to optimize LSM-tree structure, aiming to reduce the cost of processing various read/write operations. We introduce a new approach Camal, which boasts the following features: (1) ML-Aided: Camal is the first attempt to apply active learning to tune LSM-tree based key-value stores. The learning process is coupled with traditional cost models to improve the training process; (2) Decoupled Active Learning: backed by rigorous analysis, Camal adopts active learning paradigm based on a decoupled tuning of each parameter, which further accelerates the learning process; (3) Easy Extrapolation: Camal adopts an effective mechanism to incrementally update the model with the growth of the data size; (4) Dynamic Mode: Camal is able to tune LSM-tree online under dynamically changing workloads; (5) Significant System Improvement: By integrating Camal into a full system RocksDB, the system performance improves by 28% on average and up to 8x compared to a state-of-the-art RocksDB design.

The biases and discrimination of machine learning algorithms have attracted significant attention, leading to the development of various algorithms tailored to specific contexts. However, these solutions often fall short of addressing fairness issues inherent in machine learning operations. In this paper, we present a debiasing framework designed to find an optimal fair transformation of input data that maximally preserves data predictability. A distinctive feature of our approach is its flexibility and efficiency. It can be integrated with any downstream black-box classifiers, providing continuous fairness guarantees with minimal retraining efforts, even in the face of frequent data drifts, evolving fairness requirements, and batches of similar tasks. To achieve this, we leverage the normalizing flows to enable efficient, information-preserving data transformation, ensuring that no critical information is lost during the debiasing process. Additionally, we incorporate the Wasserstein distance as the unfairness measure to guide the optimization of data transformations. Finally, we introduce an efficient optimization algorithm with closed-formed gradient computations, making our framework scalable and suitable for dynamic, real-world environments.

DPO is an effective preference optimization algorithm. However, the DPO-tuned models tend to overfit on the dispreferred samples, manifested as overly long generations lacking diversity. While recent regularization approaches have endeavored to alleviate this issue by modifying the objective function, they achieved that at the cost of alignment performance degradation. In this paper, we innovatively incorporate regularization from the perspective of weight updating to curb alignment overfitting. Through the pilot experiment, we discovered that there exists a positive correlation between overfitting and the hyperspherical energy fluctuation. Hence, we introduce orthogonal finetuning for DPO via a weight-Rotated Preference Optimization (RoPO) method, which merely conducts rotational and magnitude-stretching updates on the weight parameters to maintain the hyperspherical energy invariant, thereby preserving the knowledge encoded in the angle between neurons. Extensive experiments demonstrate that our model aligns perfectly with human preferences while retaining the original expressive capacity using only 0.0086% of the trainable parameters, suggesting an effective regularization against overfitting. Specifically, RoPO outperforms DPO by up to 10 points on MT-Bench and by up to 2.8 points on AlpacaEval 2, while enhancing the generation diversity by an average of 6 points.