Core Viewpoint - The essence of humanoid robots lies in their design and the ability to learn, rather than solely relying on large models and algorithms [1][3][4]. Group 1: Importance of Design in Robotics - High Peak emphasizes that the intelligence of robots is not merely taught but designed, highlighting the need for mechanical structures that can adapt and learn from feedback [3][4]. - The current limitations of industrial robots in the 3C assembly sector stem from a lack of "soul," which refers to the absence of precise mechanisms and adaptive logic [3][7]. Group 2: Underestimated Mechanism Science - Mechanism science is identified as a critical area that is often overlooked, with the core performance of robots being determined by their mechanical structure rather than artificial intelligence [4][7]. - The traditional serial robotic arms in China face inherent flaws, limiting their application in complex tasks, while parallel and hybrid mechanisms remain underexplored due to high design difficulty [7]. Group 3: Transition from Imitation to Innovation - The shift from reverse engineering to forward design is crucial for Chinese robotics, focusing on function-driven design rather than merely replicating foreign products [8][10]. - The development of innovative mechanisms, such as the "dual screw redundant drive" for the Chang'e 5 mission, exemplifies the need for original creation in robotics [10]. Group 4: Specialization Over Generalization - High Peak argues that robots should be specialized rather than generalized, using the example of a dedicated device for precise screw fastening rather than a multifunctional robot [13][15]. - The "serial-parallel collaborative assembly robot" developed by High Peak's team demonstrates the effectiveness of specialized design in addressing industry challenges [15]. Group 5: Embodied Intelligence - The definition of robotic intelligence includes perception, decision-making, and execution as a closed loop, rather than just mimicking human actions [15][16]. - The development of the curling robot illustrates the importance of understanding the essence of tasks rather than relying on vast data training [15][16]. Group 6: Long-term Focus for Innovation - High Peak compares the slow evolution of robotic intelligence to the Stone Age, advocating for a deep focus on mechanism science to create scene experts in robotics [16].