HUGIN: Enhancing Vision-Language Planning for Autonomous Logistics Sorting

2026-08-12

AI Summary

This work addresses the challenge of joint multi-scene understanding (JMSU) across multiple camera views in autonomous parcel sorting by introducing the HUGIN training framework. HUGIN uniquely integrates endogenous data augmentation with a global contextual ranking mechanism to effectively mitigate two key issues: the scarcity of cross-scene supervision and attention dispersion caused by long visual contexts. To evaluate the approach, the authors construct SortingBench, an industrial-scale parcel sorting benchmark, on which they significantly improve the accuracy of Qwen3-VL-8B from 63.6% to 78.8%. The effectiveness and practicality of the proposed method are further validated through deployment tests on over 15,000 real-world parcels.

Abstract

Autonomous logistics sorting systems (ALSS) are an important industrial application of embodied AI, which requires joint planning over spatially disjoint camera views. We formulate this setting as Joint Multi-Scene Understanding (JMSU). With open-world visual understanding and task-planning capabilities, vision-language models (VLMs) are promising candidates for JMSU. However, directly applying existing VLMs to JMSU is non-trivial due to scarce cross-scene supervision and attention dispersion caused by long visual context in JMSU. To address these challenges, we propose HUGIN, a training framework with two complementary components. Endogenous Data Augmentation recombines verified atomic facts under operating constraints, while Global Context Ranking aligns the instruction representation more strongly with the complete visual context than with a partial visual context. To support ongoing research, we construct a high-quality industrial sorting dataset and benchmark named SortingBench from four layouts of autonomous logistics sorting systems. Across five open VLMs, HUGIN consistently outperforms matched baselines; for example, the accuracy on SortingBench of Qwen3-VL-8B increases from 63.6% to 78.8%. Additional experiments verify the effectiveness of each component and JMSU's spillover benefits in embodied tasks. Deployment tests involving more than 15,000 packages support the practical viability of VLM-based planning for autonomous logistics sorting.

Problem

• autonomous logistics sorting
• joint multi-scene understanding
• vision-language models
• spatially disjoint camera views
• task planning

Innovation

• Joint Multi-Scene Understanding
• Vision-Language Models
• Endogenous Data Augmentation
• Global Context Ranking
• Autonomous Logistics Sorting