Max Robotics — Global Service Robot Directory + US Certification

About

Hengbo Ma is a prominent researcher specializing in trajectory prediction, multi-agent behavior modeling, and autonomous systems, with significant contributions to the intersection of deep learning and intelligent mobile robotics. His work addresses one of the most critical challenges in autonomous driving and social robotics: accurately forecasting the future motion of dynamic agents in complex, interactive environments. Ma's most influential contribution, "Conditional Generative Neural System for Probabilistic Trajectory Prediction" (2019), has garnered 185 citations and established a foundational framework for probabilistic motion forecasting using generative models. He subsequently advanced the field through graph-based architectures, including his Wasserstein Graph Double-Attention Network and Spectral Temporal Graph Neural Network, demonstrating a consistent focus on capturing spatial and temporal dependencies among multiple interacting agents. His 2021 work on spatio-temporal dual-attention networks (94 citations) further solidified his expertise in scalable multi-agent prediction and tracking. Beyond prediction, Ma has explored continual learning for adaptive behavior modeling, knowledge distillation for diverse human motion forecasting, and reinforcement-guided attention mechanisms. His most recent work, "SkillDiffuser," ventures into hierarchical robotic planning using diffusion models, reflecting his evolving research vision. Collectively, his publications demonstrate a rigorous and expansive research program that has meaningfully shaped how intelligent systems perceive and anticipate the world around them.