Rising Star in AI Research: Zihan Li on SLAM, Financial AI, and Cross-Domain Innovation

——A conversation with Northeastern University researcher Zihan Li about his prolific contributions to robotics, finance, and machine learning

Zihan Li exemplifies the new generation of AI researchers who seamlessly bridge multiple technical domains. With 291 citations, an h-index of 9, and 12 published papers since 2020, this M.S. graduate from Northeastern University's Computer Science program has made significant contributions to robotics navigation, financial technology, cloud computing, and computer vision. His research portfolio demonstrates exceptional breadth while maintaining technical depth—a rare combination in today's increasingly specialized academic landscape.

Interviewer: Your research spans an unusually broad range—from SLAM technology to financial AI to cloud computing. What drives this interdisciplinary approach?

Zihan Li: I believe the most interesting problems in AI exist at the intersection of different domains. The fundamental challenges—processing multimodal data, making predictions under uncertainty, building robust systems—appear across many applications. By working in diverse areas, I can identify common patterns and transfer insights between fields.

My research naturally organizes into several themes. In robotics, I've contributed to SLAM technology and autonomous navigation systems. In financial AI, I've worked on credit risk modeling, investment decision-making, and anti-money laundering frameworks. I've also addressed practical challenges in distributed systems and explored computer vision applications.

Interviewer: Your first-author paper on robot navigation using SLAM technology has received 31 citations. Can you explain this research and its significance?

Zihan Li: This work addresses fundamental challenges in autonomous robotics—how robots can navigate and understand their environment without prior maps. SLAM (Simultaneous Localization and Mapping) is crucial for applications ranging from autonomous vehicles to rescue robots to warehouse automation.

We developed novel approaches to make SLAM more efficient and accurate for robot navigation and map construction. The key innovations involve better sensor data integration and more robust algorithms for handling dynamic environments. The citations reflect the practical importance of these problems—many research groups are building on our work for their own robotics applications.

I also contributed to research on integrating artificial intelligence with SLAM for navigation in unknown environments, which received 36 citations. By incorporating deep learning, we can make these systems more robust and adaptive. Traditional SLAM algorithms face challenges in complex, dynamic environments, but neural networks can improve localization accuracy and handle sensor noise better.

Interviewer: You've made significant contributions to financial AI. How did you transition from robotics to finance?

Zihan Li: The transition was more natural than it might seem. Both domains involve making decisions under uncertainty using complex, multimodal data. In robotics, you're processing sensor data to navigate physical space; in finance, you're processing market data to navigate decision space.

My work on bank credit risk early warning models using machine learning decision trees received 40 citations and addresses a critical need in financial institutions. We developed frameworks that can identify potential defaults earlier and more accurately than traditional approaches.

Our research on AI implementation in Chinese A-share markets, also with 35 citations, demonstrates how machine learning can enhance investment strategies. This is particularly relevant given the complexity and scale of modern financial markets. I also contributed to research on generative AI-based financial robot advisors, exploring how large language models can provide personalized investment guidance.

Additionally, my work on AI-enhanced frameworks for anti-money laundering with 24 citations shows how AI can identify suspicious patterns in cross-border transactions that traditional rule-based systems might miss.

Interviewer: Your work on cueing flight object trajectory prediction using SLAM technology received 40 citations. Can you discuss this research?

Zihan Li: This research applies SLAM principles to an unconventional domain—flight object trajectory prediction and safety assessment. The core insight is that SLAM's strength in real-time position estimation and environmental mapping can be adapted for tracking and predicting flight paths.

The high citation count reflects growing interest in using proven robotics techniques for aviation safety applications. This is another example of how cross-domain thinking can lead to innovative solutions.

Interviewer: Looking at your 2024 and 2025 publications, what are your current research priorities?

Zihan Li: I'm particularly excited about several recent directions. My work on predicting participation behavior in online collaborative learning uses large language models for text analysis and has already received 10 citations in 2025. This represents an interesting application of foundation models to educational technology.

Our 2025 paper on attention-based multimodal emotion recognition for Instagram ad engagement prediction explores how AI can understand user emotional responses to visual content. This has applications in advertising, user experience design, and human-computer interaction.

I'm also continuing work on distributed systems optimization, including traffic flow monitoring in cloud computing environments, which addresses increasingly important infrastructure challenges as systems scale.

Interviewer: With 291 total citations and an h-index of 9, your work clearly makes an impact. What drives these citation numbers?

Zihan Li: I'm grateful that other researchers find my work useful. I think several factors contribute. First, my papers tackle practical challenges that many researchers and practitioners face. When you solve problems others are working on, they naturally reference your solutions.

Second, we work hard to make our papers accessible and reproducible. Clear writing and thorough methodology sections help others build on our work. Third, many of my papers address emerging areas—like integrating LLMs with domain applications or deploying AI in financial systems—where there's significant current interest.

The i10-index of 9 means that 9 of my papers have received at least 10 citations each, which suggests consistent quality across my portfolio.

Interviewer: How did your Master's program at Northeastern University shape your research trajectory?

Zihan Li: Northeastern's Computer Science program provided excellent foundations in both theory and practice. Strong coursework in machine learning, distributed systems, and algorithms gave me the tools to tackle complex problems. The program emphasized rigorous research methodology while maintaining connections to industry applications.

Working with diverse research groups exposed me to different problem-solving approaches and application domains. Understanding how research translates to real-world applications has been invaluable for choosing impactful research directions.

Interviewer: What excites you most about the future of AI research?

Zihan Li: Several trends particularly interest me. Foundation models show impressive capabilities, but adapting them effectively to specialized domains—robotics, finance, scientific computing—remains both challenging and important.

I'm also interested in multi-agent systems. As AI systems become more capable, coordinating multiple AI agents to solve complex tasks collaboratively becomes crucial. This connects to my SLAM work on multi-robot coordination.

Finally, robust and reliable AI is essential. Moving AI from research settings to production systems requires better approaches to handling edge cases, ensuring safety, and providing interpretable decisions.

Interviewer: Thank you for sharing your insights. Your work demonstrates impressive breadth and impact across multiple AI domains.

Zihan Li: Thank you. I'm excited to continue exploring how AI can address real-world challenges across different domains. The rapid advancement in foundation models, combined with increasing computational resources, opens new possibilities. My goal is to contribute to both the theoretical understanding of AI systems and their practical deployment in ways that create meaningful value.