The Great March 100: 100 Detail-oriented Tasks for Evaluating Embodied AI Agents

Recently, with the rapid development of robot learning and imitation learning, numerous datasets and methods have emerged. However, these datasets and their task designs often lack systematic consideration and principles. This raises important questions: Do the current datasets and task designs truly advance the capabilities of robotic agents? Do evaluations on a few common tasks accurately reflect the differentiated performance of various methods proposed by different teams and evaluated on different tasks? To address these issues, we introduce the Great March 100 (\textbf{GM-100}) as the first step towards a robot learning Olympics. GM-100 consists of 100 carefully designed tasks that cover a wide range of interactions and long-tail behaviors, aiming to provide a diverse and challenging set of tasks to comprehensively evaluate the capabilities of robotic agents and promote diversity and complexity in robot dataset task designs. These tasks are developed through systematic analysis and expansion of existing task designs, combined with insights from human-object interaction primitives and object affordances. We collect a large amount of trajectory data on different robotic platforms and evaluate several baseline models. Experimental results demonstrate that the GM-100 tasks are 1) feasible to execute and 2) sufficiently challenging to effectively differentiate the performance of current VLA models. Our data and code are available at https://rhos.ai/research/gm-100.

Coordinated Pandemic Control with Large Language Model Agents as Policymaking Assistants

Effective pandemic control requires timely and coordinated policymaking across administrative regions that are intrinsically interdependent. However, human-driven responses are often fragmented and reactive, with policies formulated in isolation and adjusted only after outbreaks escalate, undermining proactive intervention and global pandemic mitigation. To address this challenge, here we propose a large language model (LLM) multi-agent policymaking framework that supports coordinated and proactive pandemic control across regions. Within our framework, each administrative region is assigned an LLM agent as an AI policymaking assistant. The agent reasons over region-specific epidemiological dynamics while communicating with other agents to account for cross-regional interdependencies. By integrating real-world data, a pandemic evolution simulator, and structured inter-agent communication, our framework enables agents to jointly explore counterfactual intervention scenarios and synthesize coordinated policy decisions through a closed-loop simulation process. We validate the proposed framework using state-level COVID-19 data from the United States between April and December 2020, together with real-world mobility records and observed policy interventions. Compared with real-world pandemic outcomes, our approach reduces cumulative infections and deaths by up to 63.7% and 40.1%, respectively, at the individual state level, and by 39.0% and 27.0%, respectively, when aggregated across states. These results demonstrate that LLM multi-agent systems can enable more effective pandemic control with coordinated policymaking...

Search Self-play: Pushing the Frontier of Agent Capability without Supervision

Reinforcement learning with verifiable rewards (RLVR) has become the mainstream technique for training LLM agents. However, RLVR highly depends on well-crafted task queries and corresponding ground-truth answers to provide accurate rewards, which requires massive human efforts and hinders the RL scaling processes, especially under agentic scenarios. Although a few recent works explore task synthesis methods, the difficulty of generated agentic tasks can hardly be controlled to provide effective RL training advantages. To achieve agentic RLVR with higher scalability, we explore self-play training for deep search agents, in which the learning LLM utilizes multi-turn search engine calling and acts simultaneously as both a task proposer and a problem solver. The task proposer aims to generate deep search queries with well-defined ground-truth answers and increasing task difficulty. The problem solver tries to handle the generated search queries and output the correct answer predictions. To ensure that each generated search query has accurate ground truth, we collect all the searching results from the proposer's trajectory as external knowledge, then conduct retrieval-augmentation generation (RAG) to test whether the proposed query can be correctly answered with all necessary search documents provided. In this search self-play (SSP) game, the proposer and the solver co-evolve their agent capabilities through both competition and cooperation. With substantial experimental results, we find that SSP can significantly improve search agents' performance uniformly on various benchmarks without any supervision under both from-scratch and continuous RL training setups. The code is at https://github.com/Alibaba-Quark/SSP.

A Physics-informed End-to-End Occupancy Framework for Motion Planning of Autonomous Vehicles

Accurate and interpretable motion planning is essential for autonomous vehicles (AVs) navigating complex and uncertain environments. While recent end-to-end occupancy prediction methods have improved environmental understanding, they typically lack explicit physical constraints, limiting safety and generalization. In this paper, we propose a unified end-to-end framework that integrates verifiable physical rules into the occupancy learning process. Specifically, we embed artificial potential fields (APF) as physics-informed guidance during network training to ensure that predicted occupancy maps are both data-efficient and physically plausible. Our architecture combines convolutional and recurrent neural networks to capture spatial and temporal dependencies while preserving model flexibility. Experimental results demonstrate that our method improves task completion rate, safety margins, and planning efficiency across diverse driving scenarios, confirming its potential for reliable deployment in real-world AV systems.

Cross-cultural Deployment of Autonomous Vehicles Using Data-light Inverse Reinforcement Learning

More than the adherence to specific traffic regulations, driving culture touches upon a more implicit part - an informal, conventional, collective behavioral pattern followed by drivers - that varies across countries, regions, and even cities. Such cultural divergence has become one of the biggest challenges in deploying autonomous vehicles (AVs) across diverse regions today. The current emergence of data-driven methods has shown a potential solution to enable culture-compatible driving through learning from data, but what if some underdeveloped regions cannot provide sufficient local data to inform driving culture? This issue is particularly significant for a broader global AV market. Here, we propose a cross-cultural deployment scheme for AVs, called data-light inverse reinforcement learning, designed to re-calibrate culture-specific AVs and assimilate them into other cultures. First, we report the divergence in driving cultures through a comprehensive comparative analysis of naturalistic driving datasets on highways from three countries: Germany, China, and the USA. Then, we demonstrate the effectiveness of our scheme by testing the expeditious cross-cultural deployment across these three countries, with cumulative testing mileage of over 56084 km. The performance is particularly advantageous when cross-cultural deployment is carried out without affluent local data. Results show that we can reduce the dependence on local data by a margin of 98.67% at best. This study is expected to bring a broader, fairer AV global market, particularly in those regions that lack enough local data to develop culture-compatible AVs.

Deployment-friendly Lane-changing Intention Prediction Powered by Brain-inspired Spiking Neural Networks

Accurate and real-time prediction of surrounding vehicles' lane-changing intentions is a critical challenge in deploying safe and efficient autonomous driving systems in open-world scenarios. Existing high-performing methods remain hard to deploy due to their high computational cost, long training times, and excessive memory requirements. Here, we propose an efficient lane-changing intention prediction approach based on brain-inspired Spiking Neural Networks (SNN). By leveraging the event-driven nature of SNN, the proposed approach enables us to encode the vehicle's states in a more efficient manner. Comparison experiments conducted on HighD and NGSIM datasets demonstrate that our method significantly improves training efficiency and reduces deployment costs while maintaining comparable prediction accuracy. Particularly, compared to the baseline, our approach reduces training time by 75% and memory usage by 99.9%. These results validate the efficiency and reliability of our method in lane-changing predictions, highlighting its potential for safe and efficient autonomous driving systems while offering significant advantages in deployment, including reduced training time, lower memory usage, and faster inference.

Automating Traffic Model Enhancement with AI Research Agent

Developing efficient traffic models is essential for optimizing transportation systems, yet current approaches remain time-intensive and susceptible to human errors due to their reliance on manual processes. Traditional workflows involve exhaustive literature reviews, formula optimization, and iterative testing, leading to inefficiencies in research. In response, we introduce the Traffic Research Agent (TR-Agent), an AI-driven system designed to autonomously develop and refine traffic models through an iterative, closed-loop process. Specifically, we divide the research pipeline into four key stages: idea generation, theory formulation, theory evaluation, and iterative optimization; and construct TR-Agent with four corresponding modules: Idea Generator, Code Generator, Evaluator, and Analyzer. Working in synergy, these modules retrieve knowledge from external resources, generate novel ideas, implement and debug models, and finally assess them on the evaluation datasets. Furthermore, the system continuously refines these models based on iterative feedback, enhancing research efficiency and model performance. Experimental results demonstrate that TR-Agent achieves significant performance improvements across multiple traffic models, including the Intelligent Driver Model (IDM) for car following, the MOBIL lane-changing model, and the Lighthill-Whitham-Richards (LWR) traffic flow model. Additionally, TR-Agent provides detailed explanations for its optimizations, allowing researchers to verify and build upon its improvements easily. This flexibility makes the framework a powerful tool for researchers in transportation and beyond. To further support research and collaboration, we have open-sourced both the code and data used in our experiments, facilitating broader access and enabling continued advancements in the field.

EcoFollower: An Environment-Friendly Car Following Model Considering Fuel Consumption

To alleviate energy shortages and environmental impacts caused by transportation, this study introduces EcoFollower, a novel eco-car-following model developed using reinforcement learning (RL) to optimize fuel consumption in car-following scenarios. Employing the NGSIM datasets, the performance of EcoFollower was assessed in comparison with the well-established Intelligent Driver Model (IDM). The findings demonstrate that EcoFollower excels in simulating realistic driving behaviors, maintaining smooth vehicle operations, and closely matching the ground truth metrics of time-to-collision (TTC), headway, and comfort. Notably, the model achieved a significant reduction in fuel consumption, lowering it by 10.42\% compared to actual driving scenarios. These results underscore the capability of RL-based models like EcoFollower to enhance autonomous vehicle algorithms, promoting safer and more energy-efficient driving strategies.

Enhance Planning with Physics-informed Safety Controllor for End-to-end Autonomous Driving

Recent years have seen a growing research interest in applications of Deep Neural Networks (DNN) on autonomous vehicle technology. The trend started with perception and prediction a few years ago and it is gradually being applied to motion planning tasks. Despite the performance of networks improve over time, DNN planners inherit the natural drawbacks of Deep Learning. Learning-based planners have limitations in achieving perfect accuracy on the training dataset and network performance can be affected by out-of-distribution problem. In this paper, we propose FusionAssurance, a novel trajectory-based end-to-end driving fusion framework which combines physics-informed control for safety assurance. By incorporating Potential Field into Model Predictive Control, FusionAssurance is capable of navigating through scenarios that are not included in the training dataset and scenarios where neural network fail to generalize. The effectiveness of the approach is demonstrated by extensive experiments under various scenarios on the CARLA benchmark.

BEVGPT: Generative Pre-trained Large Model for Autonomous Driving Prediction, Decision-Making, and Planning

Prediction, decision-making, and motion planning are essential for autonomous driving. In most contemporary works, they are considered as individual modules or combined into a multi-task learning paradigm with a shared backbone but separate task heads. However, we argue that they should be integrated into a comprehensive framework. Although several recent approaches follow this scheme, they suffer from complicated input representations and redundant framework designs. More importantly, they can not make long-term predictions about future driving scenarios. To address these issues, we rethink the necessity of each module in an autonomous driving task and incorporate only the required modules into a minimalist autonomous driving framework. We propose BEVGPT, a generative pre-trained large model that integrates driving scenario prediction, decision-making, and motion planning. The model takes the bird's-eye-view (BEV) images as the only input source and makes driving decisions based on surrounding traffic scenarios. To ensure driving trajectory feasibility and smoothness, we develop an optimization-based motion planning method. We instantiate BEVGPT on Lyft Level 5 Dataset and use Woven Planet L5Kit for realistic driving simulation. The effectiveness and robustness of the proposed framework are verified by the fact that it outperforms previous methods in 100% decision-making metrics and 66% motion planning metrics. Furthermore, the ability of our framework to accurately generate BEV images over the long term is demonstrated through the task of driving scenario prediction. To the best of our knowledge, this is the first generative pre-trained large model for autonomous driving prediction, decision-making, and motion planning with only BEV images as input.