Robotics firm X Square Robot has revealed significant progress in its pursuit of versatile household humanoids, underlined by a $100 million Series A+ funding round. The company introduced its new proprietary open-source foundation model, Wall-OSS, alongside the launch of its latest humanoid, Quanta X2. The announcement reflects growing market interest in bridging the gap between AI research and practical robotic deployments at home and in service industries. As competition heightens for leadership in embodied artificial intelligence, X Square Robot positions itself as a proactive player by making its AI model publicly accessible and targeting widespread adoption. Anticipation surrounding humanoid service robots continues to attract investors, with new models promising extended utility beyond current limitations in manipulation and decision-making.
Recent industry developments have seen robotics companies prioritizing multimodal AI and more generalized training to address real-world unpredictability. Earlier reports noted substantial investments in humanoid platforms, but these efforts were often hindered by over-specialized skill sets and reliance on limited datasets. Compared to previous attempts that prioritized bipedal mobility or single-task execution, X Square Robot’s approach integrates richer data sets and multi-layered reasoning for greater flexibility. While other companies have unveiled similar open models or platforms, Wall-OSS’s combination of vision, action, and language with large-scale training sets marks a step further in tackling real-world complexity.
How Does Wall-OSS Aim to Overcome Robotic Limitations?
Wall-OSS was developed to address persistent challenges in robotic intelligence, such as “catastrophic forgetting” where past learnings are replaced by new data, and challenges in aligning visual, linguistic, and action inputs. Equipped with a training set comprising both actual robot action data and video-augmented scenarios, Wall-OSS is designed to work across diverse robot architectures. The model enables adaptive behavior in unpredictable environments and continuous learning capabilities, purportedly making robots capable of autonomously handling varied, non-routine tasks. According to X Square Robot,
“Our embodied brain is smart enough to not only independently handle unpredictable, non-routine real-world physical settings, but also continuously learn on the go.”
What Distinguishes the Training Method Used for Wall-OSS?
A core feature of Wall-OSS is its three-stage training process, combining high-level task planning, detailed motor control, and integration of both skill sets for seamless execution. The design incorporates a shared-attention and feed-forward network, allowing vision, language, and motor pathways to operate in parallel, each optimized for specific aspects of environmental interaction. Developers can access Wall-OSS for both proprietary and third-party robots, including integration into the newly launched Quanta X2 humanoid, which merges advanced software with sophisticated mechanical hands and arms.
“These specialized pathways then work together, enabling the robot to complete complex or unfamiliar tasks more effectively,”
X Square Robot added.
Will Investment Spur Adoption of Humanoid Robots in Households?
Backed by investors such as Alibaba Cloud and HongShan, the infusion of new capital is set to drive broader implementation of Wall-OSS and development of subsequent Quanta models. The Quanta X2, equipped with a modular tool clamp and the capacity for multi-purpose end effectors, is engineered for functions in household, service, and industrial contexts. Its full-body teleoperation and dexterous hand technology target the nuanced actions required in domestic environments. Investment will also support the expansion of community-grown datasets and collaborative industry projects, further scaling accessibility to embodied AI frameworks for global developers.
The broader humanoid robotics sector is projected to accelerate sharply, with forecasts indicating market growth to $6.5 billion by the decade’s end. Current strategies revolve around deploying multifunctional, adaptable robots capable of continuous learning and generalized problem-solving. X Square Robot’s decision to open-source Wall-OSS, combine it with dexterous hardware in Quanta X2, and invite developer participation could catalyze faster transition from prototype stages to everyday utility. While several competitor platforms pursue similar ambitions, the effectiveness of real-world deployments will depend on the actual performance of training models and the ability to integrate complex sensory feedback.
Robotics developers and investors face complex challenges as robots move from labs to living rooms. For users or organizations looking to implement humanoids, considering access to reliable, open-source models like Wall-OSS, and the integration of advanced physical capabilities as seen in Quanta X2, is key. As more companies invest in dataset diversity, robust AI frameworks, and modular hardware, the coming years may determine which design philosophies best address the realities of service and household environments. Ongoing collaboration with developers and community input may prove essential as the sector seeks practical, sustainable solutions for humanoid adoption at scale.
