Goldman Sachs: Under the wave of humanoid robots, visual perception systems are undergoing an arms race

With the improvement of artificial intelligence algorithms, the enhancement of product readiness, and the expansion of end applications, the humanoid robot market is continuously growing. Goldman Sachs, in its latest research report, focuses on a key area under the wave of humanoid robots—visual perception systems. Goldman Sachs analyst Jacqueline Du pointed out in the report: The explosive growth of humanoid robots is inseparable from the evolution of their "eyes." To achieve precise perception of the surrounding environment, humanoid robots need to rely on 3D vision technology, such as stereo cameras, ToF cameras, and LiDAR. This not only provides camera suppliers with an opportunity to break out of the fiercely competitive consumer electronics market but also gives rise to the demand for new features such as night vision and anti-collision, which is expected to drive up the average selling price (ASP) of products. The visual systems of humanoid robots are diverse, with each manufacturer having its own "secret recipe." For example, Tesla's Optimus uses a relatively simple three-camera system. In contrast, UBTech's Walker S1, Fourier's GR-1, and Figure's 02 have installed up to six cameras on the robot's head and torso. Boston Dynamics' Atlas uses ToF depth sensing technology, while Unitree's G1 and AgiBot's A2 are equipped with LiDAR. Goldman Sachs noted: Although pure RGB camera solutions have higher algorithm requirements, they are currently the most effective visual perception systems. The use of ToF cameras or LiDAR can achieve more precise, stable, and reliable perception. In addition, Goldman Sachs analysts predict that by 2027, global shipments of humanoid robots will reach 76,000 units, soaring to 1.38 million units by 2035. This indicates a huge market opportunity, with the total addressable market (TAM) expected to reach $38 billion by 2030