LRAB Chapter 194 Joint Research and Development Laboratory
Chapter 194 Joint Research and Development Laboratory
One month after the licensing agreement was signed, the 402-Star Joint Laboratory was officially unveiled.
The lab is located on the third floor of Building B, the newly built headquarters of 402, covering an area of 1,200 square meters and divided into three areas. The hardware area is handled by Xingchen Technology, which houses their complete set of neural signal acquisition equipment transported from Shenzhen. The software area is handled by 402, where a small cluster of more than a dozen high-performance servers is running. In the middle is a co-working area, with each team occupying half of the workstations.
The laboratory is named "Interstellar Neuron," taking one character each from "Star" and "402." Li Mingyang was present at the unveiling ceremony; this was the second time Zuo Cheng had seen him.
Li Mingyang, dressed in the same grey Zhongshan suit as before, stood in front of the red silk cloth, holding an iron shovel tied with a red flower. Zuo Cheng stood beside him, and together they pulled the ribbon, revealing a bronze plaque on the wall.
Interstellar Neural Joint Laboratory. These eight characters are followed by the establishment date and the full names of both organizations.
Reporters snapped photos, flashes going off everywhere. Facing the camera, Li Mingyang said, "The collaboration between Xingchen Technology and 402 is a model of collaborative innovation among Chinese technology companies in cutting-edge fields. We believe that through the complementary strengths of both parties, we can accelerate the clinical translation of brain-computer interface technology."
Zuo Cheng said that 402 will do its best to make this technology available to those in need as soon as possible.
After the ceremony, Li Mingyang didn't linger and returned to Shenzhen with Zhao Wenbo. Before leaving, he told Zuo Cheng that the second-generation chip for the NX-30 was already under development, and if the joint laboratory progressed smoothly, the next-generation chip could be prioritized for testing by 402.
Zuo Cheng said he looks forward to cooperation.
The lab quickly got on track. The hardware team consisted of five people, all from Xingchen Technology, led by Chen Minghui, a man in his forties with a PhD in neuroengineering who had worked at Xingchen for eight years. The software team consisted of seven people, all from 402, personally led by Shen Yiming. The medical advisory team consisted of three people from Huaxia University School of Medicine, led by Professor Zheng, the deputy director of neurosurgery.
The first joint symposium was scheduled for the third day after the laboratory was established. Zuo Cheng personally chaired it, and there was only one topic: what should be the first research direction?
Chen Minghui said, "From a hardware perspective, I suggest optimizing the signal quality first. The NX-30's acquisition capabilities are already very strong, but there's still room for improvement in noise suppression. If the signal-to-noise ratio can be increased by another three decibels, the decoding accuracy will be significantly improved."
Shen Yiming said, "From a software perspective, I suggest starting with an adaptive decoding model. Each person's neural signal characteristics are different, and the accuracy of a general model is limited. If we can achieve rapid adaptation for each individual, the clinical value will be much greater."
Professor Zheng said, "From a medical perspective, I suggest first determining the indications. Is the primary focus on motor function reconstruction or sensory function recovery? The technical approaches and clinical pathways for these two are completely different."
After listening, Zuo Cheng said that all three suggestions made sense. His view was to first address signal quality, then adaptive decoding, and finally determine the indications. Signal quality is fundamental; without a solid foundation, everything else is just a castle in the air.
Chen Minghui nodded and said, "I agree. Signal quality optimization requires both hardware and software cooperation; we can start working on it now."
Shen Yiming asked, "How exactly do we do it?"
Chen Minghui explained that the NX-30's electrode array has 1024 channels, but not all channels have the same signal quality. Some channels are closer to nerve fibers and have strong signals, while others are off-center and have weak signals. We need to find a way to evaluate the signal quality of each channel in real time and dynamically adjust the weights.
Shen Yiming's eyes lit up. He said, "This can be done using machine learning. We can train a classifier to automatically identify channels with good signal quality and channels with poor signal quality, and then give higher weights to the good channels."
Chen Minghui said that this was exactly the idea. However, training the classifier requires labeled data, and labeling the data requires manual judgment of which channels have good signal quality, which is a huge amount of work.
Zuo Cheng said that the medical advisory team would handle the data labeling. Professor Zheng, could you arrange for graduate students to do this?
Professor Zheng said, "That's fine. Several doctoral students in our department are looking for research topics, and this is a perfect fit."
Zuo Cheng said, "Then it's settled. The first project is real-time signal quality assessment and dynamic weight optimization. Chen Minghui will be responsible for hardware signal acquisition, Shen Yiming for algorithm development, and Professor Zheng for data annotation. The first prototype will be ready within three months."
After the meeting, Zuo Cheng took a walk around the lab. Engineers in the hardware area were debugging equipment, programmers in the software area were setting up development environments, and several researchers in the co-working area were discussing technical details.
Yu Ying walked over and asked, "How are you feeling?"
Zuo Cheng said it went more smoothly than expected. Both teams were highly professional, and the communication costs were lower than I had imagined.
Yu Ying asked, "What about Tang Ning?"
Zuo Cheng said she was responsible for organizing and coordinating the technical documentation, and she did a good job. Xingchen Technology released more data than we expected; some underlying parameters weren't even mentioned in their published papers.
Yu Ying said that Li Mingyang genuinely wanted to cooperate.
Zuo Cheng said, "Yes. He wants our AI capabilities, and we want his hardware platform. It's a win-win situation. This kind of cooperation is the most stable."
The two walked to the window and looked at the busy people in the laboratory. Outside the window was Hangzhou's science and technology park, and in the distance, a corner of West Lake could be seen.
Yu Ying asked, "Do you think the first project will be successful?"
Zuo Cheng said that there were no technical problems. Signal quality assessment is a relatively mature problem, and there are readily available machine learning methods. The real challenge lies in the later stages: adaptive decoding and clinical applications.
Yu Ying asked, "What would it mean if we succeeded?"
Zuo Cheng said this means we can start recruiting volunteers and enter the clinical trial phase. It means that the eighth branch has made substantial progress.
Yu Ying glanced at him and said, "You always measure progress by the number of branches."
Zuo Cheng smiled and said, "This is my secret indicator; I can't tell you its specific meaning. But you can understand it this way: with each additional branch, the capability boundary of 402 expands outward by one ring."
Yu Ying asked, "What happens after those twelve branches?"
Zuo Cheng was silent for a while, then said, "Beyond the twelve branches lies the true starry sea."
He didn't continue, and Yu Ying didn't press him. The two stood by the window, watching the researchers coming and going in the laboratory, each lost in their own thoughts.
As night fell, the lights in the laboratory remained bright. The first project of the Interstellar Neural Joint Laboratory was thus quietly launched in the Hangzhou night.