Gasgoo Munich- The smart EV industry is racing into embodied intelligence, making humanoid robots the new battleground where automakers and tech giants are scrambling for a foothold.
On May 19, RoboParty, a fully open-source bipedal humanoid robot company, recently closed an Angel+ round worth tens of millions of dollars. Shunwei Capital led the round, with Xiaomi Strategic Investment adding to its stake. This marks Xiaomi Strategic Investment's third injection into the team in under a month, signaling that industrial capital is doubling down on the open-source path.
Image Credit: RoboParty
RoboParty positions itself as a full-stack open-source player covering algorithms, control, and hardware, aiming to bring bipedal humanoid robots to developers, universities worldwide, top-tier laboratories, and real-world industrial settings. With this fresh capital, supply chain and global ecosystem resources from Shunwei and Xiaomi will accelerate its iterations in structural design, joint modules, and scaled delivery.
On the technical front, the team took third place globally in the ICRA Reasoning to Action competition using its RPVLA solution. The event tests the closed-loop capabilities spanning semantic understanding, task reasoning, and action execution, alongside Sim2Real generalization performance — validating RoboParty's full-stack strength in perception, control, and hardware-software coordination.
The open-source ecosystem is taking off in parallel. The company's GitHub project has garnered 1,700 stars, while its developer community has surpassed 5,000 members. RoboParty also swept three major industry awards: the EAI OpenSource Top 10 Projects, the Yuanqi Award for Best Embodied Intelligence Open Source Contribution, and the LeadeRobot Humanoid Robot Scenario Pioneer Award.
Commercialization is gaining momentum. The first batch of hundreds of RPO prototypes has entered delivery and repurchase verification, with universities and laboratories worldwide proactively assembling units, reproducing movements, and debugging algorithms. Many have already pre-ordered the next-generation RP1.
