Papers
40
Total Citations
466
H-Index
13
About
Juan Rojas is a robotics researcher whose work spans intelligent manipulation, robot autonomy, and human-robot interaction — areas at the frontier of making robots more capable, safe, and collaborative in real-world environments. He has made significant contributions to robot introspection, developing methods that allow robots to detect, classify, and recover from failures during autonomous tasks using multimodal sensing and probabilistic models, including Bayesian nonparametric hidden Markov models and wrench-based action grammars. His research on continuum robotic manipulators has advanced safe control strategies through shielded multiagent reinforcement learning, particularly relevant to minimally invasive surgical applications (47 citations). Rojas has also pioneered work in human-robot collaboration, proposing LSTM-RNN-based frameworks for responsive motion generation during handovers, and contributed to scene understanding through visual-semantic graph attention networks for human-object interaction detection. Earlier foundational work on snap assembly sensing and adaptive grippers (42 citations) reflects his long-standing interest in dexterous robotic assembly. With research spanning over a decade and accumulating hundreds of citations, Rojas has established himself as a versatile and impactful contributor to the fields of autonomous manipulation and intelligent robotics.
Research Focus
Key Achievements
Top Papers
- 1
- 2Proposal of a shape adaptive gripper for robotic assembly tasks42 citations · 2016
- 3
- 4Collaborative Human-Robot Motion Generation Using LSTM-RNN29 citations · 2018
- 5
- 6A 3D object detection and pose estimation pipeline using RGB-D images22 citations · 2017
- 7Online robot introspection via wrench-based action grammars21 citations · 2017
- 8
- 9
- 10Towards snap sensing16 citations · 2013