The 2026 Asia Logistics Technology Exhibition opened in Shanghai, with Seer Intelligent and Sainade showcasing AI robot logistics applications

Today, the 2026 Asia International Logistics Technology and Transportation Systems Exhibition opened at the Shanghai New International Expo Centre, where many Shanghai companies showcased application scenarios of embodied intelligent robots in the logistics industry. Most of them are not humanoid, but AI models have become standard equipment for this type of robot. Through big-data training, the intelligence level of AI models continues to improve, enabling industrial robots to handle complex scenarios and perform tasks autonomously and flexibly. Seer Intelligent, China's first listed "robot brain" company on the Hong Kong stock market, is a Shanghai enterprise that, under the theme "Robot Brain, Making Intelligent Robots Accessible to All," showcased the operating scenarios of multiple robots: a humanoid robot selected drinks for visitors, delivery robots navigated through narrow passages and avoided obstacles, and omnidirectional ground-type forklifts moved laterally flexibly and picked accurately among different types of pallets. An omnidirectional ground-type forklift performs a working demonstration. In the robot dynamic zone, two delivery robots mixed in the same area, completing dynamic obstacle avoidance, dynamic detouring, narrow-passage crossing, and extreme S-curve passage. It is understood that real factory passages are not always spacious and regular; what the multi-robot mixed operation demonstrates is precisely the robots' ability to adapt to complex passage environments and to move from single-robot autonomous operation to multi-robot collaborative operation, and it also reflects the company's system capability for unified scheduling and management of hundreds of robots. The omnidirectional ground-type forklift demonstrated its ability to autonomously identify and accurately pick up pallets of different types, different placement orientations, and different positions. They can autonomously identify pallet targets and their poses, and complete tasks by combining positioning and motion control. For different types of pallets, there is no need to affix identification markers to them in advance, nor is it required that pallets always remain in fixed placement positions. "These robots have different forms, but behind them is the same underlying capability of the robot brain," said Ding Xia, vice president of Seer Intelligent. The robot brain adopts a trinity architecture of "model + cerebellum + nerves": the model is responsible for understanding tasks, processing information, and making decisions; the cerebellum is responsible for translating decisions into stable, coordinated actions; and the nerves connect the brain and the body, transmitting instructions to all parts of the robot body and continuously feeding real status back. This architecture is not bound to any single robot form. Through a unified software and hardware system and adaptation capabilities, humanoid robots, robot dogs, composite robots, intelligent forklifts, and others can all connect to the robot brain capability system. At present, the Seer robot brain has been adapted to more than 400 types of core components and supports more than 2,000 robot SKUs (stock keeping units), directing them to complete movement, handling, and more complex operational tasks. A humanoid robot selects drinks for visitors. Entering the Sainade exhibition area, a "real-scene loading and unloading" demonstration was underway: inside a container restored to a real scene, facing disordered cartons of different sizes, an autonomous loading and unloading robot was working methodically. It has "keen eyes" and a "brain," and can accurately identify the position and pose of each box, grasp it steadily, and place it on the conveyor line. The entire process requires no manual intervention, demonstrating the intelligent leap of the logistics "last 20 meters." A relevant technical director at Sainade introduced that loading and unloading is one of the hardest links in the intelligent transformation of industrial logistics. In real warehousing and logistics, "perfect" regular cargo stacks are only a minority; more enterprises face the "imperfect" norm of mixed-category loading, and stacks tilting or collapsing after transportation. Traditional solutions often can only handle standardized boxes of a single SKU and are often ill-suited to non-standard working conditions. Now, empowered by vertical-domain AI models, robots have predictive capability and can identify the risk of cargo collapse in advance, dynamically adjust grasping strategies and end-effector poses, and minimize the risk of collision and collapse. From "seeing" to "foreseeing," this leap not only solves the industry's chronic problem of "multi-SKU mixed palletizing," but also greatly reduces enterprises' on-site commissioning costs and reliance on manual labor. An autonomous loading and unloading robot works methodically. While achieving "unloading reliably," this Shanghai enterprise has also achieved "loading stably" through an integrated loading and unloading AI solution. In truck-loading scenarios, where carriage space is limited and cargo specifications vary, the integrated loading and unloading AI solution can generate the optimal stack pattern according to priorities such as efficiency and loading rate set by the customer. During the loading process, the system perceives the carriage status in real time and dynamically plans placement positions, tackling the hard nut of truck-loading operations. Behind these technical solutions is the R&D logic of "scenarios forcing technology." From overseas logistics warehouses to the supply chain production lines of leading domestic enterprises, vertical-domain models continue to learn and iterate amid real-world "mixed cargo stacking and packaging deformation." Every grasping data point feeds back into the algorithms and models, forming a positive flywheel of "real-scene training—data accumulation—product iteration—large-scale deployment."