Keju Peng

National University of Defense Technology

Papers

Total Citations

H-Index

About

Keju Peng is a roboticist whose research focuses on enabling autonomous navigation and environmental perception for mobile robots operating in indoor settings. His core contributions lie in visual simultaneous localization and mapping (SLAM) and robust orientation estimation using RGB-D cameras. In his highly cited 2019 work, "RGB-D SLAM Using Point–Plane Constraints for Indoor Environments," Peng introduced a novel method that fuses point and plane features to simultaneously estimate robot poses and reconstruct dense maps, directly addressing the fundamental challenge of pose estimation and map reconstruction for autonomous behavior. This paper has garnered 19 citations, underscoring its influence in the field. Complementing this, his 2019 paper "Robust Visual Compass Using Hybrid Features for Indoor Environments" proposed a drift-free visual compass that leverages hybrid features—including planes and lines—to provide reliable orientation estimates for motion control and 3D mapping. With 13 citations, this work highlights his innovative approach to solving orientation estimation, a critical component for stable navigation. Peng’s research is particularly notable for its practical application of geometric constraints to improve robustness in cluttered, texture-poor indoor environments, making his contributions valuable for advancing real-world robotic autonomy.