Publications detail
The Great March 100: 100 Detail-oriented Tasks for Evaluating Embodied AI Agents
Abstract
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 (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.
Ziyu Wang, Chenyuan Liu, Yushun Xiang, Runhao Zhang, Yu Zhang, et al. “The Great March 100: 100 Detail-oriented Tasks for Evaluating Embodied AI Agents.” arXiv:2601.11421, 2026.
01
Benchmarking as editing
This publication behaves more like an editorial benchmark system than a new model paper. Its value comes from giving scattered tasks, metrics, and failure patterns a shared shell for comparison.
Once the work moves into VLA, the meaningful differences often appear only after supervision is reduced and environmental feedback becomes messier.
02
Evaluation choices
The benchmark is divided into demonstration scarcity, drifting language constraints, visual obstruction, and delayed long-horizon feedback.
That split makes the resulting notes easier to interpret later than a single aggregated score.
03
Current status
The work is still closer to a benchmark draft than a finished paper, but it already has a publication-ready rhythm: question, method, observation, next step.
Future iterations will add clearer data-split policy and more explicit failure visualizations.