MAROKO133 Update ai: Which Agent Causes Task Failures and When?Researchers from PSU and Du

📌 MAROKO133 Breaking ai: Which Agent Causes Task Failures and When?Researchers fro

Share My Research is Synced’s column that welcomes scholars to share their own research breakthroughs with over 1.5M global AI enthusiasts. Beyond technological advances, Share My Research also calls for interesting stories behind the research and exciting research ideas. Contact us: [email protected]

Meet the authors
Institutions: Penn State University, Duke University, Google DeepMind, University of Washington, Meta, Nanyang Technological University, and Oregon State University. The co-first authors are Shaokun Zhang of Penn State University and Ming Yin of Duke University.

In recent years, LLM Multi-Agent systems have garnered widespread attention for their collaborative approach to solving complex problems. However, it’s a common scenario for these systems to fail at a task despite a flurry of activity. This leaves developers with a critical question: which agent, at what point, was responsible for the failure? Sifting through vast interaction logs to pinpoint the root cause feels like finding a needle in a haystack—a time-consuming and labor-intensive effort.
 
This is a familiar frustration for developers. In increasingly complex Multi-Agent systems, failures are not only common but also incredibly difficult to diagnose due to the autonomous nature of agent collaboration and long information chains. Without a way to quickly identify the source of a failure, system iteration and optimization grind to a halt.
 
To address this challenge, researchers from Penn State University and Duke University, in collaboration with institutions including Google DeepMind, have introduced the novel research problem of “Automated Failure Attribution.” They have constructed the first benchmark dataset for this task, Who&When, and have developed and evaluated several automated attribution methods. This work not only highlights the complexity of the task but also paves a new path toward enhancing the reliability of LLM Multi-Agent systems.

The paper has been accepted as a Spotlight presentation at the top-tier machine learning conference, ICML 2025, and the code and dataset are now fully open-source.

Paper：https://arxiv.org/pdf/2505.00212
Code：https://github.com/mingyin1/Agents_Failure_Attribution
Dataset：https://huggingface.co/datasets/Kevin355/Who_and_When
 
 
Research Background and Challenges
LLM-driven Multi-Agent systems have demonstrated immense potential across many domains. However, these systems are fragile; errors by a single agent, misunderstandings between agents, or mistakes in information transmission can lead to the failure of the entire task.

Currently, when a system fails, developers are often left with manual and inefficient methods for debugging:
Manual Log Archaeology : Developers must manually review lengthy interaction logs to find the source of the problem.
Reliance on Expertise : The debugging process is highly dependent on the developer’s deep understanding of the system and the task at hand.
 
This “needle in a haystack” approach to debugging is not only inefficient but also severely hinders rapid system iteration and the improvement of system reliability. There is an urgent need for an automated, systematic method to pinpoint the cause of failures, effectively bridging the gap between “evaluation results” and “system improvement.”

Core Contributions
This paper makes several groundbreaking contributions to address the challenges above:
1. Defining a New Problem: The paper is the first to formalize “automated failure attribution” as a specific research task. This task is defined by identifying the failure-responsible agent and the decisive error step that led to the task’s failure.
2. Constructing the First Benchmark Dataset: Who&When : This dataset includes a wide range of failure logs collected from 127 LLM Multi-Agent systems, which were either algorithmically generated or hand-crafted by experts to ensure realism and diversity. Each failure log is accompanied by fine-grained human annotations for:
Who: The agent responsible for the failure.
When: The specific interaction step where the decisive error occurred.
Why: A natural language explanation of the cause of the failure.

3. Exploring Initial “Automated Attribution” Methods : Using the Who&When dataset, the paper designs and assesses three distinct methods for automated failure attribution:
– All-at-Once: This method provides the LLM with the user query and the complete failure log, asking it to identify the responsible agent and the decisive error step in a single pass. While cost-effective, it may struggle to pinpoint precise errors in long contexts.
– Step-by-Step: This approach mimics manual debugging by having the LLM review the interaction log sequentially, making a judgment at each step until the error is found. It is more precise at locating the error step but incurs higher costs and risks accumulating errors.
– Binary Search: A compromise between the first two methods, this strategy repeatedly divides the log in half, using the LLM to determine which segment contains the error. It then recursively searches the identified segment, offering a balance of cost and performance.

Experimental Results and Key Findings 
Experiments were conducted in two settings: one where the LLM knows the ground truth answer to the problem the Multi-Agent system is trying to solve (With Ground Truth) and one where it does not (Without Ground Truth). The primary model used was GPT-4o, though other models were also tested. The systematic evaluation of these methods on the Who&When dataset yielded several important insights:
– A Long Way to Go: Current methods are far from perfect. Even the best-performing single method achieved an accuracy of only about 53.5% in identifying the responsible agent and a mere 14.2% in pinpointing the exact error step. Some methods performed even worse than random guessing, underscoring the difficulty of the task.
– No “All-in-One” Solution: Different methods excel at different aspects of the problem. The All-at-Once method is better at identifying “Who,” while the Step-by-Step method is more effective at determining “When.” The Binary Search method provides a middle-ground performance.
 

– Hybrid Approaches Show Promise but at a High Cost: The researchers found that combining different methods, such as using the All-at-Once approach to identify a potential agent and then applying the Step-by-Step method to find the error, can improve overall performance. However, this comes with a significant increase in computational cost.

– State-of-the-Art Models Struggle: Surprisingly, even the most advanced reasoning models, like OpenAI o1 and DeepSeek R1, find this task challenging.- This h…

Konten dipersingkat otomatis.

🔗 Sumber: syncedreview.com

📌 MAROKO133 Hot ai: Affordable humanoid robot kit at $15,000 pushes advanced robot

The race to build humanoid robots is moving beyond secretive corporate labs and into the hands of independent developers.

Singapore-based Menlo Research has unveiled a DIY version of its open-source humanoid robot, Asimov, aimed at hobbyists, researchers, and robotics enthusiasts.

Priced at around $15,000—close to the project’s estimated bill-of-materials cost—the kit reflects a broader push to make bipedal robotics more accessible.

Recently, a hobbyist created a life-size sci-fi droid replica using 3D printing and AI voice technology, showcasing affordable tools for interactive home robotics and automation.

Modular humanoid platform

Menlo Research’s open-source humanoid robot kit has a strong focus on modular engineering and simulation-driven robotics development.

The 3.93 feet (1.20 meter) tall humanoid weighs around 77 pounds (35 kilograms and features more than 25 degrees of freedom, offering builders a fully customizable research platform rather than a consumer-ready robot. Delivered completely unassembled, the system includes detailed manuals and instructional build videos aimed at developers and advanced hobbyists.

A major technical highlight is the robot’s modular architecture. Independent leg, arm, torso, and head sections connect through universal motor mounting fixtures, allowing users to swap or upgrade components without redesigning the entire platform. The approach reduces maintenance complexity while enabling rapid experimentation with new actuators and control systems, reports Humanoids Daily (HD).

The humanoid also incorporates a parallel Revolute-Spherical-Universal (RSU) ankle mechanism that provides two degrees of freedom for roll and pitch movement. The design improves torque distribution across the ankle joint and allows the robot to respond more naturally to uneven terrain and ground reaction forces during walking.

To simplify locomotion control, Asimov uses passive articulated toes rather than powered toe actuators. These non-actuated joints assist with the transition from stance to push-off, improving traction and balance while reducing computational overhead and mechanical complexity.

Most structural components are optimized for Multi Jet Fusion (MJF) 3D printing, enabling the production of strong, lightweight parts without relying on expensive CNC machining processes. This lowers manufacturing costs while making replacement and customization easier for developers, according to reports.

Realistic training system

Asimov’s software stack is built around a “Processor-in-the-Loop” (PIL) simulation approach that deliberately moves away from idealized robotics models. Instead of assuming clean, perfectly timed sensor data and deterministic physics, the training environment injects realistic operational imperfections to better mirror real-world conditions.

This includes simulated CANBus communication delays of up to 9 milliseconds, producing stale or out-of-sync control signals, as well as artificially generated sensor noise through an I2C emulation layer. These disturbances are designed to replicate the unpredictability and latency inherent in physical robot systems.

At the learning core, the system uses an Asymmetric Actor-Critic reinforcement learning framework. The “critic” network is granted access to privileged ground-truth simulation data, enabling accurate evaluation of state and reward signals. In contrast, the “actor” operates under constrained conditions, receiving only noisy, delayed sensor inputs similar to what onboard hardware would experience.

By training under this mismatch, the policy learns to tolerate uncertainty and partial observability. The result is zero-shot sim-to-real transfer, allowing the robot to walk forwards, backwards, and recover from external pushes directly on hardware without additional tuning or calibration, reports HD.

The kit isn’t inexpensive, with a target price of around $15,000. However, Asimov publishes a full bill of materials on its GitHub repository, allowing builders to source components independently and potentially reduce costs. According to Hackaday, while still a significant investment, it is considered far more accessible than earlier humanoid robotics systems that required millions in development funding.

🔗 Sumber: interestingengineering.com