At ZNAI, we are committed to advancing the frontiers of artificial intelligence to address real-world challenges in autonomous systems and sustainable urban development. Our research is grounded in the belief that intelligent technologies must not only be powerful and accurate but also environmentally sustainable, socially inclusive, and ethically responsible.
1. AI in 3D Vision for Autonomous Navigation
One of our core research thrusts focuses on the development of AI algorithms for 3D vision in the context of autonomous navigation. We investigate robust perception and mapping systems that integrate multi-modal sensor data—including stereo vision, LiDAR point clouds, and depth-aware RGB-D inputs—to build accurate and efficient 3D representations of urban environments. These models are foundational to navigation tasks such as obstacle avoidance, semantic localization, and trajectory planning in complex, dynamic settings.
Our work emphasizes accuracy, generalization, and sustainability. To this end, we explore lightweight neural architectures and transformer-based fusion techniques that reduce energy consumption during both training and inference, without compromising on performance. We also study self-supervised learning for 3D scene understanding, which reduces the reliance on labeled data and enhances scalability. Particular attention is given to edge-AI deployment scenarios where computational constraints are strict, such as in assistive navigation tools for visually impaired individuals.
2. Decision-Making, Urban Sustainability, and Reinforcement Learning
In parallel, ZNAI is actively engaged in research on AI-enabled decision-making systems to support urban sustainability. We leverage reinforcement learning (RL) and multi-agent systems to design adaptive control policies for applications including traffic signal optimization, pedestrian flow regulation, and dynamic public transport routing. These RL-based solutions are augmented with graph neural networks to encode spatial relationships in urban layouts and with causal inference methods to ensure transparency and interpretability in decisions.
A significant strand of our research addresses how AI can contribute to inclusive mobility and climate-aware planning. We develop simulation environments to model and evaluate long-term impacts of mobility policies, incorporating factors such as emissions, accessibility equity, and energy demand. Our models are designed to support participatory planning processes and are being extended with human-in-the-loop capabilities to align algorithmic outputs with community needs and values.
3. Cross-Cutting Themes: Accessibility, Ethics, and Education
Our work integrates ethical considerations from the ground up. Accessibility—particularly for visually disabled individuals—is not treated as an afterthought, but as a central design criterion. This includes the co-design of AI-powered assistive navigation systems that operate reliably in GPS-denied environments and support context-aware feedback through auditory or haptic interfaces.
Beyond research, ZNAI is committed to education and public engagement.
