- Industrial embodied AI firm bets on “world model” for loading automation
- System simulates object deformation and stack stability before robot action
Industrial embodied AI robotics company Senad (赛那德) has raised 300 million yuan ($44 million) in a Series C round of funding, as it pushes deeper into commercial deployment of logistics automation systems.
The round was led by by digital freight platform Full Truck Alliance, with participation from Haitong Capital, Huayi Ventures and two Oriza Holdings-managed funds.
Existing shareholders including Fortune Capital, Long Capital and Prime Light Capital also increased their stakes, according to the company.
Dual-core architecture
The funding comes as Senad debuts what it calls the world’s “first physical engine” tailored to vertical in-warehouse cargo loading scenarios.
Founded in 2012 and headquartered in Zhejiang’s Jiaxing, Senad said the capital will be used to accelerate mass production and delivery of its iLoabot series of embodied loading robots, as well as R&D on its dual-core architecture combining an “industrial brain” and a “robotic cerebellum.”
Alongside the funding announcement, the company launched Senad Robot Insight-World V3.0, a world model designed specifically for logistics unloading and loading environments.
Compensating for VLA’s shortcomings
Unlike conventional vision-language-action (VLA) systems that react to what they see in real time, Senad’s model introduces spatial-temporal prediction, trained on large-scale real-world interaction data.
The system allows robots to simulate how cargo stacks will deform and shift before making a grasp.
In traditional loading scenarios, robots typically decide grasping actions based on a single camera frame. But when boxes are stacked, weight pressure can cause deformation or instability, increasing the risk of collapse or damage if the robot misjudges force or angle.
Senad’s system instead “rehearses” physical outcomes in advance—predicting how stacks will shift under different grasping strategies and selecting the safest and most stable approach before execution.
Predictive machines
The company said this shift effectively turns loading robots from reactive machines into predictive systems, improving performance in unstructured environments and significantly enhancing success rates and operational safety.
Senad said its iLoabot-M autonomous unloading robot has achieved a 100% success rate in structured environments and around 80% in certain unstructured scenarios.
Through a closed-loop system of real-world data capture and virtual model iteration, the company has also accumulated a rare dataset on cargo shapes and stacking patterns, it added.
Logistics and FMCG customers
Senad’s robots are now deployed at scale in sectors including tobacco, beverages and pharmaceuticals, with customers such as JD.com, SF Express and snack brand Three Squirrels.
The company said it is the only embodied loading robotics firm in China with proven commercial deployment at scale.
Its product portfolio includes autonomous loading, heavy-duty loading, bulk unloading and heavy-load humanoid systems covering end-to-end logistics scenarios.
It has also exported to more than 20 countries and regions in North America, Japan, South Korea and Western Europe.
