SEEN·E Robotics / 深诣机器人
About
About SEEN·E Robotics / 深诣机器人
SEEN·E Robotics combines embodied-AI models with robot hardware for industrial and logistics work. Its SE-1 foundation model is presented as a system for understanding tasks, predicting physical outcomes and generating actions. The company describes a World Model MoE architecture, experience-matching methods and asynchronous off-policy reinforcement learning to improve policies using demonstrations, historical data and interaction feedback. Its logistics platform Sculptor targets mixed-case palletizing, picking and handling; Edge is a dual-arm platform for fine industrial manipulation and interactive tasks.
Founded in 2026, SEEN·E is led by Geek+ co-founder Kai Liu, with technical co-founders Yu Luo and Kangyao Huang and chief scientist Huaping Liu. It made a public appearance at WRC in August 2026 and presented SE-1, Sculptor and Edge at CIIF in October. Demonstrations included 0.1 mm-class cable insertion, mixed-box palletizing and interactive candy packing. The platforms are demonstrated in exhibition and research settings. The company’s research organization also publishes manipulation-learning work, including PRISM with Chinese university collaborators.
Team
- Kai Liu / 刘凯 - Founder and Chief Executive Officer
- Yu Luo / 罗宇 - Technical Co-founder - Develops the company’s embodied-AI models and reinforcement-learning methods.
- Kangyao Huang / 黄康尧 - Technical Co-founder - Connects embodied-system engineering with the company’s industrial and logistics robot programme.
- Huaping Liu / 刘华平 - Chief Scientist
Timeline
Events
- 2026-10-13 - Product Launch - SEEN·E unveils SE-1 model, Sculptor and Edge robots at CIIF
Sculptor uses configurable bases and end tools for mixed-case palletizing, picking, truck handling and related logistics tasks. Edge demonstrated 0.1 mm-class cable insertion under perturbations and interactive, multi-step candy packing. The model combines a World Model MoE architecture, OMPO experience matching and asynchronous off-policy reinforcement learning.
Source - 2026-08-19 - Demonstration - SEEN·E demonstrates adaptive precision manipulation at WRC2026
The robot demonstrated 0.1 mm-class industrial insertion while adapting to changes in the operating scene. The company combines learning from demonstrations with interaction feedback to improve policies after initial task acquisition. Kai Liu’s WRC presentation addressed faster task learning and sustained skill improvement for industrial robotics.
Source
Context
Shared workflows
- Tianyuanxing (天元兴)
workflow adjacency - Industrial Embodied Control
- Palladyne AI Corp.
workflow adjacency - Industrial Embodied Control
- RLWRLD
workflow adjacency - Industrial Embodied Control
Context tags
- Industrial embodied control - Workflow - https://www.leaderobot.com/news/10189 — Edge manufacturing manipulation and Sculptor handling are defining hardware development directions.
- Vision-language-action control - Capability - https://arxiv.org/abs/2610.14704 — Company-authored SE-1/PRISM research supports embodied policy/control context; does not turn company into model-only infrastructure.