VitaBench: Benchmarking LLM Agents with Versatile Interactive Tasks in Real-world Applications

As LLM-based agents are increasingly deployed in real-life scenarios, existing benchmarks fail to capture their inherent complexity of handling extensive information, leveraging diverse resources, and managing dynamic user interactions. To address this gap, we introduce VitaBench, a challenging benchmark that evaluates agents on versatile interactive tasks grounded in real-world settings. Drawing from daily applications in food delivery, in-store consumption, and online travel services, VitaBench presents agents with the most complex life-serving simulation environment to date, comprising 66 tools. Through a framework that eliminates domain-specific policies, we enable flexible composition of these scenarios and tools, yielding 100 cross-scenario tasks (main results) and 300 single-scenario tasks. Each task is derived from multiple real user requests and requires agents to reason across temporal and spatial dimensions, utilize complex tool sets, proactively clarify ambiguous instructions, and track shifting user intent throughout multi-turn conversations. Moreover, we propose a rubric-based sliding window evaluator, enabling robust assessment of diverse solution pathways in complex environments and stochastic interactions. Our comprehensive evaluation reveals that even the most advanced models achieve only 30% success rate on cross-scenario tasks, and less than 50% success rate on others. Overall, we believe VitaBench will serve as a valuable resource for advancing the development of AI agents in practical real-world applications. The code, dataset, and leaderboard are available at https://vitabench.github.io/

STARC: See-Through-Wall Augmented Reality Framework for Human-Robot Collaboration in Emergency Response

In emergency response missions, first responders must navigate cluttered indoor environments where occlusions block direct line-of-sight, concealing both life-threatening hazards and victims in need of rescue. We present STARC, a see-through AR framework for human-robot collaboration that fuses mobile-robot mapping with responder-mounted LiDAR sensing. A ground robot running LiDAR-inertial odometry performs large-area exploration and 3D human detection, while helmet- or handheld-mounted LiDAR on the responder is registered to the robot's global map via relative pose estimation. This cross-LiDAR alignment enables consistent first-person projection of detected humans and their point clouds - rendered in AR with low latency - into the responder's view. By providing real-time visualization of hidden occupants and hazards, STARC enhances situational awareness and reduces operator risk. Experiments in simulation, lab setups, and tactical field trials confirm robust pose alignment, reliable detections, and stable overlays, underscoring the potential of our system for fire-fighting, disaster relief, and other safety-critical operations. Code and design will be open-sourced upon acceptance.

DVDP: An End-to-End Policy for Mobile Robot Visual Docking with RGB-D Perception

Automatic docking has long been a significant challenge in the field of mobile robotics. Compared to other automatic docking methods, visual docking methods offer higher precision and lower deployment costs, making them an efficient and promising choice for this task. However, visual docking methods impose strict requirements on the robot's initial position at the start of the docking process. To overcome the limitations of current vision-based methods, we propose an innovative end-to-end visual docking method named DVDP(direct visual docking policy). This approach requires only a binocular RGB-D camera installed on the mobile robot to directly output the robot's docking path, achieving end-to-end automatic docking. Furthermore, we have collected a large-scale dataset of mobile robot visual automatic docking dataset through a combination of virtual and real environments using the Unity 3D platform and actual mobile robot setups. We developed a series of evaluation metrics to quantify the performance of the end-to-end visual docking method. Extensive experiments, including benchmarks against leading perception backbones adapted into our framework, demonstrate that our method achieves superior performance. Finally, real-world deployment on the SCOUT Mini confirmed DVDP's efficacy, with our model generating smooth, feasible docking trajectories that meet physical constraints and reach the target pose.

HCRide: Harmonizing Passenger Fairness and Driver Preference for Human-Centered Ride-Hailing

Order dispatch systems play a vital role in ride-hailing services, which directly influence operator revenue, driver profit, and passenger experience. Most existing work focuses on improving system efficiency in terms of operator revenue, which may cause a bad experience for both passengers and drivers. Hence, in this work, we aim to design a human-centered ride-hailing system by considering both passenger fairness and driver preference without compromising the overall system efficiency. However, it is nontrivial to achieve this target due to the potential conflicts between passenger fairness and driver preference since optimizing one may sacrifice the other. To address this challenge, we design HCRide, a Human-Centered Ride-hailing system based on a novel multi-agent reinforcement learning algorithm called Harmonization-oriented Actor-Bi-Critic (Habic), which includes three major components (i.e., a multi-agent competition mechanism, a dynamic Actor network, and a Bi-Critic network) to optimize system efficiency and passenger fairness with driver preference consideration. We extensively evaluate our HCRide using two real-world ride-hailing datasets from Shenzhen and New York City. Experimental results show our HCRide effectively improves system efficiency by 2.02%, fairness by 5.39%, and driver preference by 10.21% compared to state-of-the-art baselines.

IONext: Unlocking the Next Era of Inertial Odometry

Researchers have increasingly adopted Transformer-based models for inertial odometry. While Transformers excel at modeling long-range dependencies, their limited sensitivity to local, fine-grained motion variations and lack of inherent inductive biases often hinder localization accuracy and generalization. Recent studies have shown that incorporating large-kernel convolutions and Transformer-inspired architectural designs into CNN can effectively expand the receptive field, thereby improving global motion perception. Motivated by these insights, we propose a novel CNN-based module called the Dual-wing Adaptive Dynamic Mixer (DADM), which adaptively captures both global motion patterns and local, fine-grained motion features from dynamic inputs. This module dynamically generates selective weights based on the input, enabling efficient multi-scale feature aggregation. To further improve temporal modeling, we introduce the Spatio-Temporal Gating Unit (STGU), which selectively extracts representative and task-relevant motion features in the temporal domain. This unit addresses the limitations of temporal modeling observed in existing CNN approaches. Built upon DADM and STGU, we present a new CNN-based inertial odometry backbone, named Next Era of Inertial Odometry (IONext). Extensive experiments on six public datasets demonstrate that IONext consistently outperforms state-of-the-art (SOTA) Transformer- and CNN-based methods. For instance, on the RNIN dataset, IONext reduces the average ATE by 10% and the average RTE by 12% compared to the representative model iMOT.

S$^4$C: Speculative Sampling with Syntactic and Semantic Coherence for Efficient Inference of Large Language Models

Large language models (LLMs) exhibit remarkable reasoning capabilities across diverse downstream tasks. However, their autoregressive nature leads to substantial inference latency, posing challenges for real-time applications. Speculative sampling mitigates this issue by introducing a drafting phase followed by a parallel validation phase, enabling faster token generation and verification. Existing approaches, however, overlook the inherent coherence in text generation, limiting their efficiency. To address this gap, we propose a Speculative Sampling with Syntactic and Semantic Coherence (S$^4$C) framework, which extends speculative sampling by leveraging multi-head drafting for rapid token generation and a continuous verification tree for efficient candidate validation and feature reuse. Experimental results demonstrate that S$^4$C surpasses baseline methods across mainstream tasks, offering enhanced efficiency, parallelism, and the ability to generate more valid tokens with fewer computational resources. On Spec-bench benchmarks, S$^4$C achieves an acceleration ratio of 2.26x-2.60x, outperforming state-of-the-art methods.

X-Scene: Large-Scale Driving Scene Generation with High Fidelity and Flexible Controllability

Diffusion models are advancing autonomous driving by enabling realistic data synthesis, predictive end-to-end planning, and closed-loop simulation, with a primary focus on temporally consistent generation. However, the generation of large-scale 3D scenes that require spatial coherence remains underexplored. In this paper, we propose X-Scene, a novel framework for large-scale driving scene generation that achieves both geometric intricacy and appearance fidelity, while offering flexible controllability. Specifically, X-Scene supports multi-granular control, including low-level conditions such as user-provided or text-driven layout for detailed scene composition and high-level semantic guidance such as user-intent and LLM-enriched text prompts for efficient customization. To enhance geometrical and visual fidelity, we introduce a unified pipeline that sequentially generates 3D semantic occupancy and the corresponding multiview images, while ensuring alignment between modalities. Additionally, we extend the generated local region into a large-scale scene through consistency-aware scene outpainting, which extrapolates new occupancy and images conditioned on the previously generated area, enhancing spatial continuity and preserving visual coherence. The resulting scenes are lifted into high-quality 3DGS representations, supporting diverse applications such as scene exploration. Comprehensive experiments demonstrate that X-Scene significantly advances controllability and fidelity for large-scale driving scene generation, empowering data generation and simulation for autonomous driving.

How to Provably Improve Return Conditioned Supervised Learning?

In sequential decision-making problems, Return-Conditioned Supervised Learning (RCSL) has gained increasing recognition for its simplicity and stability in modern decision-making tasks. Unlike traditional offline reinforcement learning (RL) algorithms, RCSL frames policy learning as a supervised learning problem by taking both the state and return as input. This approach eliminates the instability often associated with temporal difference (TD) learning in offline RL. However, RCSL has been criticized for lacking the stitching property, meaning its performance is inherently limited by the quality of the policy used to generate the offline dataset. To address this limitation, we propose a principled and simple framework called Reinforced RCSL. The key innovation of our framework is the introduction of a concept we call the in-distribution optimal return-to-go. This mechanism leverages our policy to identify the best achievable in-dataset future return based on the current state, avoiding the need for complex return augmentation techniques. Our theoretical analysis demonstrates that Reinforced RCSL can consistently outperform the standard RCSL approach. Empirical results further validate our claims, showing significant performance improvements across a range of benchmarks.

MOBODY: Model Based Off-Dynamics Offline Reinforcement Learning

We study the off-dynamics offline reinforcement learning problem, where the goal is to learn a policy from offline datasets collected from source and target domains with mismatched transition. Existing off-dynamics offline RL methods typically either filter source transitions that resemble those of the target domain or apply reward augmentation to source data, both constrained by the limited transitions available from the target domain. As a result, the learned policy is unable to explore target domain beyond the offline datasets. We propose MOBODY, a Model-Based Off-Dynamics offline RL algorithm that addresses this limitation by enabling exploration of the target domain via learned dynamics. MOBODY generates new synthetic transitions in the target domain through model rollouts, which are used as data augmentation during offline policy learning. Unlike existing model-based methods that learn dynamics from a single domain, MOBODY tackles the challenge of mismatched dynamics by leveraging both source and target datasets. Directly merging these datasets can bias the learned model toward source dynamics. Instead, MOBODY learns target dynamics by discovering a shared latent representation of states and transitions across domains through representation learning. To stabilize training, MOBODY incorporates a behavior cloning loss that regularizes the policy. Specifically, we introduce a Q-weighted behavior cloning loss that regularizes the policy toward actions with high target-domain Q-values, rather than uniformly imitating all actions in the dataset. These Q-values are learned from an enhanced target dataset composed of offline target data, augmented source data, and rollout data from the learned target dynamics. We evaluate MOBODY on MuJoCo benchmarks and show that it significantly outperforms state-of-the-art baselines, with especially pronounced improvements in challenging scenarios.

LLM-Guided Reinforcement Learning: Addressing Training Bottlenecks through Policy Modulation

While reinforcement learning (RL) has achieved notable success in various domains, training effective policies for complex tasks remains challenging. Agents often converge to local optima and fail to maximize long-term rewards. Existing approaches to mitigate training bottlenecks typically fall into two categories: (i) Automated policy refinement, which identifies critical states from past trajectories to guide policy updates, but suffers from costly and uncertain model training; and (ii) Human-in-the-loop refinement, where human feedback is used to correct agent behavior, but this does not scale well to environments with large or continuous action spaces. In this work, we design a large language model-guided policy modulation framework that leverages LLMs to improve RL training without additional model training or human intervention. We first prompt an LLM to identify critical states from a sub-optimal agent's trajectories. Based on these states, the LLM then provides action suggestions and assigns implicit rewards to guide policy refinement. Experiments across standard RL benchmarks demonstrate that our method outperforms state-of-the-art baselines, highlighting the effectiveness of LLM-based explanations in addressing RL training bottlenecks.