Jim Aldon D'Souza

Papers

Total Citations

H-Index

About

Jim Aldon D'Souza is a rising researcher at the forefront of autonomous systems and multi-agent perception, whose work bridges the critical gap between robust prediction and real-world sensor reliability. His primary research focuses on multi-agent trajectory prediction and multimodal scene understanding for robotics and autonomous vehicles. D'Souza’s most significant contribution is the development of **Latent Variable Sequential Set Transformers**, a novel architecture that jointly models social, temporal, and contextual information to produce socially consistent future trajectories for multiple agents. This work, published in 2021, has already garnered **38 citations**, underscoring its impact on the field of safe robotic control. He further extended this framework with **AutoBots**, which refines the sequential set transformer for more efficient multi-agent coordination. Demonstrating his versatility, D'Souza also tackles sensor limitations in adverse conditions through his work on **long-range acoustic beamforming**, showing how sound can augment or replace traditional electromagnetic sensors like lidar and cameras in low-light or challenging environments. His research is not only technically innovative but also directly addresses practical safety and reliability challenges in autonomous navigation, making him a notable voice in next-generation robotic perception.