RoboTalk: Learning Multi-Robot Communication and Coordination
Overview
RoboTalk is a synthetic dataset for training VLMs on multi-robot tasks, achieving 77% success on novel tasks through multimodal demonstrations and leader-follower protocols. It includes 7,950 trajectories across 53 kitchen tasks with natural language and tool calls. This research was published on arXiv in October 2026. Latest: RoboTalk dataset enables improved coordination in multi-robot systems through synthetic training data. (60 chars)
Generated from attributed reports · Updated 1 hours ago
Event evidence and corrections
0 attributed source owners. Ownership does not establish independent confirmation. Quantities are reported separately and are never added together.
Data origin: robot trajectories
“tion and skill-level action selection from multimodal demonstrations remains underexplored for small vision-language models (VLMs) intended for on-device deployment. To address this gap, we introduce RoboTalk, a synthetic data-generation pipeline and dataset of 7,950 multimodal trajectories spanning 53 mobile-manipulation kitchen tasks for training small VLMs to communicate and coordinate. The dataset includes a lead”
Exact source · revision 1Source owner not reported
Reported quantity · trajectories: 7950 other · trajectory count
“tion and skill-level action selection from multimodal demonstrations remains underexplored for small vision-language models (VLMs) intended for on-device deployment. To address this gap, we introduce RoboTalk, a synthetic data-generation pipeline and dataset of 7,950 multimodal trajectories spanning 53 mobile-manipulation kitchen tasks for training small VLMs to communicate and coordinate. The dataset includes a lead”
Exact source · revision 1Source owner not reported
Reported quantity · Reported success rate: 2 percent · Basis not reported
“er-follower planning protocol, tool calls (perception, manipulation, navigation, and communication), rationale traces, and diversified natural-language communication. Fine-tuning open-source models on our dataset can reach 77% success on novel held-out tasks, a significant improvement over the untuned open source models, which had a success rate of around ~2%.”
Exact source · revision 1Source owner not reported
Report timeline
Follow attributed reports and material updates.
- arXiv Robotics — research abstractsRoboTalk: Learning Multi-Robot Communication and Coordination from Multimodal Demonstrations
This research presents RoboTalk, a synthetic data-generation pipeline and dataset for training small vision-language models (VLMs) on multi-robot communication and coordination. The dataset includes 7,950 multimodal trajectories across 53 mobile-manipulation kitchen tasks, featuring leader-follower protocols, tool calls, and natural language communication. Fine-tuning open-source models on this dataset achieves 77% success on novel tasks, a significant improvement over untuned models.
Timezone · UTC
Article dates follow your selected timezone. Briefing editions use Hong Kong time (UTC+8).