Recently, 36kR learned that the general foot robot company “LimX Dynamics” completed the angel round and Pre-A round of financing, with a total amount of nearly 200 million yuan. The angel round was jointly conducted by several top venture capital institutions, including Fengrui Capital, Wisdom Capital, Mingshi Capital and Kunzhong Capital, while the Pre-A round was led by Oasis Capital and Legend Venture Capital also participated in the round. This round of financing will mainly be used for the landing of four-wheeled foot products, humanoid robots and embodied intelligence key technologies.
Founded in 2022 and headquartered in Shenzhen, Zhuji Power focuses on making universal foot robots. At present, the company has released four-wheeled robot products, and is also doing technical research and pre-research on humanoid robots.
In September 2023, Yuji Power released a four-wheeled robot, combining the leg and wheel structure, equipped with a number of high-performance joints, and launched a unique wheel-footed product, which can realize real-time terrain perception and all-terrain mobility, covering extremely difficult discrete terrain such as going up and down stairs.
According to Zhang Wei, founder of Zuji Power, the current foot robot is not mature in terms of terrain adaptation ability, most foot robots are in the “weak perception” stage, and the mobile efficiency is low, which is not conducive to the landing of products.
The four-wheel foot robot can improve the mobility of the foot robot, and the wheeled movement is mainly adopted in the flat land. In the complex terrain, the leg structure can help the robot improve the terrain adaptability and realize the mobility on the all-terrain road surface, including the ground, lawn, stairs and so on. The moving speed of the four-legged robot is 3 or 4 times faster than that of the conventional four-legged robot, and the walking distance is longer under the same unit energy consumption.
The product is planned to be used in industrial inspection, logistics distribution, special operations, scientific research and education and other to B scenarios, and is expected to begin accepting reservations in the fourth quarter of 2023.
“In our communication with many customers, we found that the current quadruped robot can not meet the requirements of customers for terrain and efficiency in landing applications, and we are choosing characteristic and representative industry customers to create benchmark landing cases together.” “Zhang Wei said.
The actual landing of the foot robot in the scene not only requires hardware with high reliability and stability, but also needs to have terrain-based perception ability, so that the hardware can be used in actual environments such as factories. “Weak perception” is also a bottleneck for most current legged robots.
The core team has more than ten years of profound accumulation in the field of robotics, and Yuji Power has the world’s leading perception-based motion control algorithm, foot robot reinforcement learning algorithm and self-developed high-performance foot joint. This allowed them to effectively develop a practical foot robot product. In addition to the foot robot, embodied intelligence is also the field of international dynamics to maintain high attention.
After the popularity of large models, humanoid robots are considered to be the most promising scenarios for carrying large models. Large language models have the ability to talk, which can give humanoid robots a general brain, so that they can smoothly talk to the outside world, but also increase the “decision” ability of task understanding, splitting and logical reasoning.
Embodied intelligence There are a lot of players pouring in, some from the perspective of large models to empower robot brains, and some companies from the perspective of motion control. The company hopes to develop humanoid robots with perception-based motion control and perception-based operation capabilities.
Zhang Wei, the founder of Zhuji Power, has been researching humanoid robots for many years. As early as 2021, he and his team at Ohio State University took the lead in the United States in using reinforcement learning to complete the stable walking of humanoid biped robots indoors and outdoors. After the establishment of the international power, further, through the model driven and AI hybrid way to promote the research and development of biped robots, is the industry’s few team to closely combine the two.
Zhang Wei has the actual research and development experience of humanoid robots, but the strategy of Zhuji power for humanoid robots is to strive to maintain the technical pre-research ability and leading advantage in the field, and not rush to launch products.
“The technology of humanoid robot products is not yet mature, and there are still many key technologies that have not been completed. It is not so difficult to launch a humanoid robot that can walk a few steps on the flat ground. Our humanoid biped robot is to have the ability to move all terrain.” Zhang Wei told 36KR.
The idea of the humanoid robot is to first develop a stable and reliable humanoid robot that can move on complex terrain, as well as basic operation capabilities, and complete the technical breakthrough and software framework construction of the humanoid robot; Then overcome the functional requirements required when the humanoid robot lands. A humanoid biped robot with progressive power is expected to be launched in 2023.
Cost and landing scenarios are the core issues that humanoid robots will face in the next step. Humanoid robots are still in the early embryonic stage of the circuit, and it is difficult for the industry to determine a commercial closed loop.
In Zhang Wei’s view, humanoid robots can be used in closed scenes, but there is still a long way to calculate the economic account, which does not depend on individual companies, but depends on the entire technical and academic progress.
Dr. Zhang Wei, founder of Zhuji Power, is a professor at Southern University of Science and Technology. Prior to the establishment of Zhuji Power, he was a professor at Ohio State University in the United States, where he studied control theory and application, robotics, and machine learning for a long time. At present, the company has more than 40 employees.