Skip to content
Back to selected work

Dec 2024 · Humanoid Robotics / Conversational AI / Multimodal (Vision/Voice/NLP) / Agibot A2 / Enterprise Delivery

Bumblebee: Multimodal AI, Embodied

A two-meter humanoid that brings vision, voice and bilingual dialogue into one coherent interaction experience. Built on Agibot A2 with our own conversational AI stack, Bumblebee combines multimodal understanding with consistent character expression. We led AI orchestration, robot integration and on-site deployment for Volcano Engine’s flagship exhibition hall.

Bumblebee: Multimodal AI, Embodied

The challenge

A public-facing humanoid needs vision, voice and dialogue to feel like one experience. The task was to deliver bilingual conversation and consistent character expression for a fixed exhibition opening.

The approach

Our conversational AI stack coordinated vision, speech and language processing on the two-meter Agibot A2 platform. Work covered AI orchestration, robot integration, character behavior and on-site deployment.

The outcome

Delivered to Volcano Engine’s flagship exhibition hall on schedule, bringing bilingual, in-character interaction into a public-facing humanoid installation.

Let’s talk research and engineering