In multi-robot systems, although the idea of behaviors allows for an efficient solution to low-level tasks, high-level missions can rarely be achieved by execution a single behavior. contrast this, sequence would provide requisite expressiveness, but there are no priori guarantees that is composable in sense robots actually execute it. order guarantee provably correct composition behaviors, Finite-Time Convergence Control Barrier Functions introduced this paper terminal configuration one...

Large language models (LLMs) have shown the potential to be integrated into human daily lives. Therefore, user preference is most critical criterion for assessing LLMs' performance in real-world scenarios. However, existing benchmarks mainly focus on measuring models' accuracy using multi-choice questions, which limits understanding of their capabilities real applications. We fill this gap by proposing a comprehensive Chinese benchmark SuperCLUE, named after another popular LLM CLUE....

Considerable research efforts have been devoted to ensuring that large language models (LLMs) align with human values and generate safe text. However, an excessive focus on sensitivity certain topics can compromise the model's robustness in following instructions, thereby impacting its overall performance completing tasks. Previous benchmarks for jailbreaking LLMs primarily focused evaluating safety of without considering their robustness. In this paper, we propose a benchmark assesses both...

Automatically synthesizing consistent models is a key prerequisite for many testing scenarios in autonomous driving or software tool validation where model-based systems engineering techniques are frequently used to ensure designated coverage of critical cornercases. From practical perspective, an inconsistent model irrelevant as test case (e.g. false positive), thus each synthetic needs simultaneously satisfy various structural and attribute well-formedness constraints. While different...

We consider the general problem of moving a large number networked robots toward goal position through cluttered environment while preserving network communication connectivity and avoiding both inter-robot collisions collision with obstacles. In contrast to previous approaches that either plan complete paths for each individual robot in high-dimensional joint configuration space or control group as whole explicit constraints on group's boundary pairwise distance, we propose novel...

Automatically synthesizing consistent models is a key prerequisite for many testing scenarios in autonomous driving to ensure designated coverage of critical corner cases. An inconsistent model irrelevant as test case (e.g., false positive); thus, each synthetic needs simultaneously satisfy various structural and attribute constraints, which includes complex geometric constraints traffic scenarios. While different logic solvers or dedicated graph have recently been developed, they fail...

In many applications, multi-robot systems are required to achieve multiple objectives. For these multi-objective tasks, it is oftentimes hard design a single control policy that fulfills all the objectives simultaneously. this paper, we focus on tasks can be decomposed into set of simple subtasks. Controllers for subtasks individually designed and then combined entire team. One significant feature our work subtask controllers along with their underlying manifolds. When controller other...

In the age of network globalization, a monotonically increasing amount traffic data is produced continuously every day. Capturing HTTP fast and accurately has become valuable issue. this paper, we analyzed Hypertext Transfer Protocol (HTTP) found out patterns its structure. Then designed acquiring system with an advanced trie-based protocol parsing algorithm. Based on trie, can extract user-defined fields without mismatch or backtracking. According to characteristics specific structure,...

Today, earnings mismanagement in China’s enterprises has become a serious problem as managers conduct financial fraud by means of management, hindering overall economic development. Upon shareholders’ requirements and investors’ concerns, should disclose real information. The essay analyzes the revenue function generated manager shareholder through an evolutionary theory model where team enterprise shareholders are both game parties. After building model, utilizes Python to stimulate...

In this paper, we study the relation between generalized inverse properties of an element in a ring with involution and related equations. Mainly, by exploring existence solution given set expressions general to these constructed equation, obtain lot new characterizations EP elements, partial isometries, SEP Hermitian elements normal elements.

Pornographic content occurring in human-machine interaction dialogues can cause severe side effects for users open-domain dialogue systems. However, research on detecting pornographic language within is an important subject that rarely studied. To advance this direction, we introduce CensorChat, a monitoring dataset aimed at whether the session contains content. end, collect real-life wild and break them down into single utterances single-turn dialogues, with last utterance spoken by...

A system of linear equations $L$ is common over $\mathbb{F}_p$ if, as $n\to\infty$, any 2-coloring $\mathbb{F}_p^n$ gives asymptotically at least many monochromatic solutions to a random 2-coloring. The notion systems analogous that graphs, i.e., graphs whose density in 2-edge-coloring cliques minimized by the coloring. Saad and Wolf initiated systematic study on identifying systems, built upon earlier work Cameron-Cilleruelo-Serra. When single equation, Fox-Pham-Zhao gave complete...

The advent of the Segment Anything Model (SAM) marks a significant milestone for interactive segmentation using generalist models. As late fusion model, SAM extracts image embeddings once and merges them with prompts in later interactions. This strategy limits models ability to extract detailed information from prompted target zone. Current specialist utilize early that encodes combination images objects, yet repetitive complex computations on result high latency. key these issues is...

We present a novel observation about the behavior of offline reinforcement learning (RL) algorithms: on many benchmark datasets, RL can produce well-performing and safe policies even when trained with "wrong" reward labels, such as those that are zero everywhere or negatives true rewards. This phenomenon cannot be easily explained by RL's return maximization objective. Moreover, it gives degree robustness is uncharacteristic its online counterparts, which known to sensitive design....

NSFW (Not Safe for Work) content, in the context of a dialogue, can have severe side effects on users open-domain dialogue systems. However, research detecting language, especially sexually explicit within has significantly lagged behind. To address this issue, we introduce CensorChat, monitoring dataset aimed at detection. Leveraging knowledge distillation techniques involving GPT-4 and ChatGPT, offers cost-effective means constructing content detectors. The process entails collecting...

We present a technique for learning how to solve multi-robot mission that requires interaction with an external environment by observing expert system executing the same mission. define as team of robots equipped library controllers, each designed specific task, supervised policy appropriately selects controllers based on states and environment. The objective is un-trained (i.e., imitator system) but agnostic policy, execute mission, performances comparable those system. From un-annotated...

Based on the existing pivot rules, simplex method for linear programming is not polynomial in worst case. Therefore optimal of crucial. This study proposes rule to find all shortest paths problems based Monte Carlo tree search (MCTS). Specifically, we first propose SimplexPseudoTree transfer into mode while avoiding repeated basis variables. Secondly, four reinforcement learning (RL) models with two actions and rewards make suitable method. Thirdly, set a new action selection criterion...

This paper presents a theory of optimization fabrics, second-order differential equations that encode nominal behaviors on space and can be used to define the behavior smooth optimizer. Optimization fabrics commonalities among problems reflect structure itself, enabling processes intelligently navigate each problem even when optimizing simple naive potential functions. Importantly, over fabric is inherently asymptotically stable. The majority this dedicated development tool set for design...

Celebrity special collections have important academic, historical, and spiritual values. The resource types include books, letters, manuscripts, pictures, audio, videos, physical objects, etc., which need different data description standards, content management models, preservation methods possess the characteristics of heterogeneous resources across multiple fields like libraries, archives, museums, science technology museums. Compared with other collection resources, this category has more...

In many scenarios involving human interaction with a remote swarm, the operator needs to be periodically updated state information from robotic swarm. A complete representation of swarm is high dimensional and perceptually inaccessible human. Thus, summary often required. addition, it case that human-swarm communication channel extremely bandwidth constrained may have latency. This motivates need for itself compute its own transmission operator. The generated by selecting subset robots,...

Effective use of camera-based vision systems is essential for robust performance in autonomous off-road driving, particularly the high-speed regime. Despite success structured, on-road settings, current end-to-end approaches scene prediction have yet to be successfully adapted complex outdoor terrain. To this end, we present TerrainNet, a vision-based terrain perception system semantic and geometric aggressive, navigation. The approach relies on several key insights practical considerations...

We study a new paradigm for sequential decision making, called offline policy learning from observations (PLfO). Offline PLfO aims to learn policies using datasets with substandard qualities: 1) only subset of trajectories is labeled rewards, 2) may not contain actions, 3) be high quality, and 4) the data have full coverage. Such imperfection common in real-world scenarios, encompasses many existing setups, including imitation (IL), IL (ILfO), reinforcement (RL). In this work, we present...