UBTech: Fleet of humanoid robots work collaboratively at Zeekr
UBTech is training its humanoid Walker S1 robots so that they can perform tasks together in industry. The first training session took place at Zeekr.
Humanoid Walker S1 robots work collaboratively at Zeekr.
(Image: UBTech Robotics)
UBTech, developer and manufacturer of the Walker S1 humanoid robot, is the first company in the world to use several humanoid robots in a factory that have to work together to complete tasks. The so-called Training 2.0 of the Walker S1 robots is taking place in the Chinese 5G Smart Factory of car manufacturer Zeekr. The intelligent BrainNet framework is used to coordinate the collaboration of the humanoid robots.
To date, various humanoid robots have been used by car manufacturers such as BMW, Mercedes-Benz, Tesla and Nio as lone wolves. They carry out individual, repetitive tasks in production or logistics. However, many of the tasks can only be completed as part of a team, so an entity is needed to coordinate the robots' actions. It determines, for example, which robot performs which subtask and which activities may need to be carried out by two or more robots.
BrainNet
However, UBtech's BrainNet goes one step further and forms a kind of swarm intelligence system. It essentially consists of a “super brain” and an intelligent “underbrain”. The superbrain is based on a comprehensive multimodal model that enables intelligent hybrid decision-making. It manages complex tasks within a production line, for example. The intelligent lower brain, on the other hand, is based on a transformer model. It integrates general, cross-field perception and collaborative control of multiple robots. This supports parallel, distributed learning and accelerates the development and transfer of skills.
UBTech claims that BrainNet can continuously develop itself. It is capable of processing large amounts of data and equipping humanoid robots “with a human-like common sense”. This enables the Walker S1 robots to subdivide, plan and coordinate their tasks independently. As a result, the robots optimize their cooperation in complex industrial processes.
Empfohlener redaktioneller Inhalt
Mit Ihrer Zustimmung wird hier ein externes YouTube-Video (Google Ireland Limited) geladen.
Ich bin damit einverstanden, dass mir externe Inhalte angezeigt werden. Damit können personenbezogene Daten an Drittplattformen (Google Ireland Limited) übermittelt werden. Mehr dazu in unserer Datenschutzerklärung.
UBTech's multimodal thinking model was trained with an industrial data set that the robotics company has collected in practice with its humanoid Walker S robots in various car factories. The model is able to adapt to specific work functions.
Expansion to other industrial companies
UBTech tested the BrainNet in Zeekr's 5G Intelligent Factory. Several dozen Walker S1s were trained and used in complex production areas such as vehicle and final assembly, as well as quality inspection. The humanoid robots worked together on individual tasks. These included sorting and handling tasks, as well as precision assembly. They proved that humanoid robots are capable of working collaboratively in real industrial environments.
Videos by heise
UBtech hopes that Training 2.0 at Zeekr will accelerate the collaborative training of multiple humanoid robots and the development of AI models. UBTech now wants to expand the use of its Walker S1 robots to other partner factories, including the Chinese car manufacturer Geely Auto, which is the parent company of Volvo Cars, Polestar and Lotus, among others.
(olb)
