ECOVACS fully open-sources its Bajie embodied AI robot, hardware and software included

ECOVACS is laying its entire Bajie embodied intelligence robot stack open for developers. The floor-cleaning robot giant recently opened the 'Bajie Open-Source Innovation Space' in Suzhou, offering free access to both the hardware platform and underlying system source code.

The Bajie robot, launched this year, features a six-degree-of-freedom robotic arm with an 800mm reach, a silent mobile chassis, a two-finger gripper, and a UV cleaning unit. It is powered by dual RK3588 SoCs that can run a 7-billion-parameter model locally via PCI-E expansion, eliminating the need to send every decision to the cloud.

ECOVACS opened four usage pathways: non-programmers can command the AI agent through natural language, developers can use C++ or Python SDKs to control underlying capabilities, researchers can leverage SSH and serial ports for algorithm validation, and enterprises can customize the platform for commercial applications.

ECOVACS chairman Qian Dongqi explicitly rejected the humanoid approach. He argued that robots were originally designed for labor and their appearance matters less than their ability to navigate real home environments, where wheeled locomotion is quieter and more energy-efficient than bipedal walking.

On technical approach, Qian expressed skepticism about the prevailing VLA (Vision-Language-Action) end-to-end model trend, arguing that a single general model cannot cover the vast variety of home settings. He favors a personalized approach — building solid foundational capabilities first, then tailoring the robot for each household.

The initial target users are ECOVACS' existing 'pioneer user' base from its floor-cleaning robots. Through in-home research, the team defined first-use scenarios including toy organization, shoe tidying, laundry assistance, item retrieval, and desk cleaning — all concrete home-space management tasks rather than broad general-purpose robotics.

Beyond the physical facility in Suzhou, ECOVACS also launched an online community at ecovacs-tech.com where developers can remotely debug robots, browse full source code on Atomgit and GitHub, access application examples, and install skill modules without any application or approval process.

Qian acknowledged that open-sourcing is not pure philanthropy — he said that perception of and action execution in robots cannot be perfected by a single company, requiring community collaboration. ECOVACS is not chasing a 'human-like' narrative but is laying out the hardest part of embodied AI for the community to build upon, piece by piece.