“…In addition, hardware acceleration has become one of the key technologies to support real-time processing of complex robotic systems. Research in Reference [2] and Reference [3] has made significant progress in hardware-level real-time workload forensics, which provides an important reference for hardware optimization of robotic systems. At the same time, the research of Reference [4] greatly accelerated the training process of graph neural networks through extremely fast GPU kernel design, further supporting the efficiency of robots in processing large-scale graphics data.…”