Mengkai Hu

Peking University, Northwestern Polytechnical University

Papers

Total Citations

H-Index

About

Mengkai Hu is a robotics researcher whose work bridges deep learning and humanoid locomotion. His most influential contribution, "Robotic grasp detection using a novel two-stage approach" (2021, 24 citations), addresses a critical limitation in end-to-end grasp detection systems. By proposing a two-stage framework that decouples the detection process, Hu’s method reduces the strict dependency on large, high-quality training datasets—a common bottleneck in CNN-based robotic manipulation. This innovation makes grasp detection more practical and adaptable for real-world applications, where data is often scarce or noisy. Earlier, Hu explored humanoid robotics with his work on "Humanoid robot's omnidirectional walking" (2015, 5 citations). In this study, he developed an omnidirectional gait generation model using linear centroid motion and cubic spline interpolation, grounded in the ZMP (Zero Moment Point) stability criterion. This approach enables humanoid robots to walk smoothly in any direction, enhancing their versatility and efficiency. Though his citation counts are modest, Hu’s contributions are notable for their practical focus—tackling real-world constraints in both manipulation and locomotion. His work is particularly relevant for researchers seeking robust, data-efficient solutions in robotic grasping and stable, flexible gait generation for humanoid platforms.