A decade ago, mobile specialized robots in China were merely small-scale "experimental products" cautiously tested in chemical plants and refineries, often squeezed between pipelines, tank areas, and high-voltage grids. Today, Tianchuang Robotics has deployed explosion-proof humanoid robots in these environments. From mobile specialized inspection robots to embodied AI and explosion-proof humanoids, Tianchuang has spent nearly a decade refining these machines. These robots now inspect, monitor gauges, and detect anomalies in chemical plants, power stations, coal mines, and boiler rooms—places too hazardous for prolonged human presence.
Two pivotal moments solidified Liu Shuang's decision to pivot: the leap in reasoning and interaction capabilities brought by AI large models, and Elon Musk's introduction of Optimus, which reimagined the possibilities of the humanoid robotics industry. In Liu's view, the robotics industry can be divided into three eras: pre-2015, dominated by traditional industrial robots; 2015–2023, defined by collaborative, service, transport, and inspection robots; and post-2023, where consensus has converged on embodied AI and humanoid/anthromorphic robots.
Tianchuang's response to this era is to develop an explosion-proof humanoid robot capable of real "work" in highly dangerous, reliability-obsessed environments—rather than competing on price or mass production. Unlike many robotics startups that begin with exhibitions, funding, or demos, Tianchuang's story started in a chemical plant's night shift, a coal mine's ventilation shaft, a 70°C oil and gas facility, and missions where robots had to replace humans in perilous tasks. This is the company's core ethos: deploying robotic intelligence where humans cannot go.
With the rise of embodied AI and humanoids showcased on platforms like Tesla and China's Spring Festival Gala, the industry is reevaluating these once-niche "specialized robots." In a dialogue with Sina Finance, Liu Shuang elaborated on Tianchuang's strategic shift from specialized robots to embodied AI, addressing key questions:
1. **Why focus on industrial scenarios rather than consumer-centric bipedal humanoids?** Liu argues that industrial settings prioritize upper-body functionality (e.g., valve-turning, heavy lifting) over bipedal locomotion. Wheeled or tracked bases suffice for most industrial mobility needs, while explosion-proofing, high payloads, and operational versatility are critical.
2. **Why caution against premature mass production?** Liu warns that the industry's rush to scale production overlooks unmet demand. For instance, humanoids in car factories are relegated to test lines because their current efficiency lags behind production-line requirements. He advocates first proving viable commercial applications before optimizing costs.
3. **How does China fare in the global robotics race?** Liu believes China leads in hardware (sensors, actuators, supply chains) but trails in foundational AI models. However, he draws parallels to EVs and solar panels, where China eventually dominated manufacturing. For embodied AI, he predicts a similar trajectory once scalable applications emerge.
Tianchuang's flagship, the explosion-proof humanoid *Tiankuai-1*, exemplifies this industrial focus. Designed for hazardous environments like chemical plants and mines, it meets stringent requirements for explosion-proofing, heavy-load handling (e.g., valve operation), and adaptability to industrial layouts. Its development leveraged Tianchuang's decade-long expertise in ruggedized robotics, including collaborations on explosion-proof designs for other industries.
Key technical challenges, such as battery thermal management in explosion-proof conditions, were addressed through innovations like sodium-ion batteries for cold environments, active cooling systems for extreme heat, and lightweight structural reinforcements. Liu emphasizes that cost is secondary to functionality: "If a robot can truly perform tasks, even at $20K, customers will pay. If it’s just a toy, $1K feels expensive."
Looking ahead, Tianchuang aims to close the loop on "actionable" robots in industrial maintenance, progressing from mobility and sensing to operational autonomy. Liu underscores the need for ecosystem-building, akin to how PLC systems entrenched brand loyalty in automation. He also highlights the importance of academia-industry collaboration, noting Tianchuang's ties to Tsinghua University's new Embodied AI and Robotics Institute.
In Liu's view, the embodied AI revolution is a "long and wide" race still in its early stages. The priority isn’t timelines or cost-cutting but steadily advancing robots' real-world capabilities—one task at a time.
(Images: T9-W explosion-proof wheeled robot inspecting a steel mill; Tiankuai-1 in a配电房; Liu Shuang at the production facility.)
Comments