Gledson Melotti

Instituto Federal do Espírito Santo, Institute for Systems Engineering and Computers

Papers

Total Citations

H-Index

About

Gledson Melotti is a researcher at the forefront of autonomous driving perception, specializing in multimodal deep learning and sensor fusion. His work focuses on integrating camera and LIDAR data to enhance object recognition and pedestrian classification for robotic vehicles. In his most-cited paper, "Multimodal Deep-Learning for Object Recognition Combining Camera and LIDAR Data" (2020, 46 citations), Melotti pioneered late fusion strategies that combine visual and range information, significantly improving detection accuracy in complex environments. He further advanced pedestrian safety with "CNN-LIDAR pedestrian classification: combining range and reflectance data" (2018, 12 citations), demonstrating how LIDAR reflectance data can refine classification in low-visibility conditions. Notably, Melotti addresses a critical flaw in autonomous systems in "Reducing Overconfidence Predictions in Autonomous Driving Perception" (2022, 11 citations), where he tackles the overconfidence problem in Softmax and Sigmoid outputs—a key issue for reliable decision-making in safety-critical applications. His contributions bridge the gap between sensor technology and robust AI, earning recognition for improving both the accuracy and trustworthiness of perception systems. With a growing citation impact, Melotti’s work continues to shape the future of autonomous robotics and intelligent transportation.