Max Robotics — Global Service Robot Directory + US Certification

About

Arnie Sen is pioneering the intersection of active perception and semantic scene understanding for autonomous robotics. His research focuses on enabling robots to intelligently interact with cluttered, dynamic environments through three core areas: active object detection, zero-shot instance segmentation, and language-guided navigation. Sen’s most influential work, “Learning to View: Decision Transformers for Active Object Detection” (2023, 16 citations), introduces a novel framework that couples planning with perception, allowing robots to strategically reposition themselves to gather more informative visual data—a paradigm shift from traditional independent perception systems. He further advances robotic vision with “SupeRGB-D: Zero-Shot Instance Segmentation in Cluttered Indoor Environments” (2023, 11 citations), which tackles the critical challenge of detecting and segmenting objects without extensive manual annotation. His most recent contributions include VLPG-Nav (2024), which integrates visual language pose graphs with object localization probability maps for precise object-centric navigation, and a groundbreaking algorithm for modeling uncertainty in 3D Gaussian Splatting through continuous semantic splatting (2025). Sen’s work is essential reading for researchers building robust, perception-driven robots capable of operating in complex human environments.