L4 autonomous delivery vehicles (RoboVans) with 3m³/6m³/12m³ cargo volumes covering multiple urban delivery scenarios; the X6 is the mainstay model for RaaS instant delivery
End-to-end foundation model; the freight industry's first vision-to-action foundation model, with a fully end-to-end architecture
Technology foundation; Chen Wei specializes in foundation model R&D - 'whoever designs the model on the chip owns its full definition'
General-purpose humanoid robot; combined front-of-house interaction and object manipulation; can work with coffee machines to prepare drinks
Dedicated delivery robots; the mainstay of restaurant food delivery, with 4 trays and automatic obstacle avoidance
Dedicated delivery robot for full-process hotel delivery; collaborates with humanoid robots on laundry scenarios
Dedicated robots covering hotels, healthcare, commercial cleaning, and other scenarios
AGV/AMR; shelf carrying; core product for e-commerce warehousing
AGV/AMR; tote transport and high-density storage; the flagship Honghu C56 achieves double-deep storage and retrieval, improving storage efficiency by 70%+
AGV; automated forklift pallet handling; the FD150A uses SLAM navigation with ±10mm positioning accuracy, suited to narrow aisles in manufacturing
AMR; customized for special scenarios such as photovoltaics, lithium batteries, and pharmaceuticals
Integrated solution; the world's first dual-robot collaboration combining dense storage and goods-to-person picking, solving the pain point of split-case picking for massive SKU counts
Software; logistics execution system (AI material allocation) and warehouse execution system (order wave grouping)
Core component. Simultaneously outputs 15 types of physical information including 6-axis force, material, temperature and rebound; 0.01 N precision, 1,000 Hz sampling frequency.
Based on multi-dimensional tactile array sensors, achieving fine grasping and manipulation.
Full-chain in-house humanoid robot with integrated tactile perception.
Covering 100,000+ interaction scenarios; part of the data open for research use.
Self-developed foundation model enabling robots to understand complex tasks and perform fine manipulation; the embodied model MiMo-Embodied is open-sourced
End-to-end multimodal embodied foundation model deeply fusing multi-sensor, LLM, VLM, VLA and world models for cross-scenario / cross-task / cross-embodiment intelligent generalization
Post-training framework; an online post-training service deployable as a specialized model with one click after training