Papers
3
Total Citations
23
H-Index
3
About
Pierre Marza is a rising researcher at the forefront of embodied AI and autonomous navigation, with a focused expertise in integrating neural implicit representations into robotic perception. His work addresses a fundamental challenge: how can an agent autonomously explore and understand a new environment without relying on pre-collected human data? Marza’s major contributions center on enabling agents to dynamically learn scene representations—such as Neural Radiance Fields (NeRF)—on the fly during navigation. His most cited paper, “Multi-Object Navigation with dynamically learned neural implicit representations” (2023, 14 citations), pioneers a method that combines end-to-end learning with implicit mapping, allowing agents to navigate and map multi-object spaces without traditional SLAM. In his follow-up work, “AutoNeRF” (2023–2024, 9 combined citations), Marza tackles the critical bottleneck of data collection for NeRF training by developing autonomous agents that actively explore and gather the necessary visual data themselves. This shift from passive, human-curated datasets to active, agent-driven exploration marks a notable achievement, positioning Marza’s research at the intersection of computer vision, robotics, and reinforcement learning. His work is particularly impactful for students and researchers interested in building truly autonomous systems that can learn and adapt in real-world, unstructured environments.
Research Focus
Key Achievements
Top Papers
- 1
- 2AutoNeRF: Training Implicit Scene Representations with Autonomous Agents5 citations · 2023
- 3AutoNeRF: Training Implicit Scene Representations with Autonomous Agents4 citations · 2024