Technische Universität München Robotics and Embedded Systems
 

Caixia Cai

 

Research Assistant

E-Mail caica@in.tum.de
Room MI 03.07.037
Phone +49.89.289.18141
Fax +49.89.289.18107
Address Institut für Informatik VI
Technische Universität München
Boltzmannstraße 3
85748 Garching bei München
Germany
Homepage http://www.i6.in.tum.de/Main/Caica

Caixia Cai
 

Curriculum Vitæ

Current Research Interests

Teaching

Demos

0. Constrained based Robot Control

In this video, a prioritized, multi-task control framework is presented, that can realize force-based tasks in systems for industrial robots. In this framework, multiple tasks and constraints can be simultaneously controlled, and different priorities for tasks are accomplished through null-space projections. The proposed framework is evaluated for several typical industrial robotics applications like grasping, welding, erasing and peg-in-hole.

1. Visual Servoing

* 6D Stereo Uncalibrated Image-Based Visual Servoing (IROS 2014)

This video shows an implementation of a 6D image-based visual servoing control. The extrinsic parameters of the camera and the kinematic/dynamic parameters of the robot are considered as unknown. The visual-servoing control uses error functions defined in a 3D visual space (measured in pixels). The control approach can be integrated seamless with other control strategies to obtain different dynamic behaviours. In order to overcome the limitation of the stereo-rig (visual occlusion), a new camera model is designed together with a robust control law. Now, in case of occlusion, the stereo-rig can be moved. The control scheme will adapt to the new camera parameters.The demo also shows different dynamic/kinematic behavior for the robot.

* 3D Stereo Uncalibrated Image-Based Visual Servoing (IROS 2013)

* Safe Human-Robot Interaction: Human Avoidance

2. Robotics

* Robot Kinematic and Dynamic Modeling

help the development of useful tools for robot modeling (Kinematic and Dynamic models)

* Human-Robot Interaction (Teaching & Execution)

During the teaching phase, the reasoning engine uses the information from the perception module to associate involved objects/actors with the task. The Process and Tasks are stored in an Ontology-based Knowledge Data Base. This knowledge can be used later in more complex Process. The Semantic abstraction allows the decoupling of the Problem Space from the Solution Space. Thereby, the problem description and analysis is isolated from the execution and scenario-specific parameters. This research is associated to the project SMErobotics .

* Human-Robot Collaboration

The architecture can be used for Human-Robot Collaboration. The semantic description allows to integrate Human tasks with Robot Tasks. In this way, the strengths of both actors can be used to optimize the process. This research is associated to the project SMErobotics .

3. Computer Vision

* Scene Perception and Recognition Using Point Clouds in Human-Robot Interaction Scenarios

For More Information please visit the Homepage : Dr. Emmanuel Carlos Dean Leon and my colleague Nikhil Somani

Publications

[1] Caixia Cai. 6D Visual Servoing for Industrial Manipulators applied to Human-Robot Interaction Scenarios. Dissertation (phd thesis), Technische Universität München, Munich, Germany, 2017. [ DOI | .bib | .pdf ]
[2] Nikhil Somani, Markus Rickert, Andre Gaschler, Caixia Cai, Alexander Perzylo, and Alois Knoll. Task level robot programming using prioritized non-linear inequality constraints. In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Daejeon, Republic of Korea, October 2016. https://youtu.be/baet9IkTK04, https://github.com/nsomani/constraint-controller-library. [ DOI | .bib | .pdf ]
[3] Caixia Cai, Nikhil Somani, and Alois Knoll. Orthogonal Image Features for Visual Servoing of a 6-DOF Manipulator with Uncalibrated Stereo Cameras. IEEE transactions on Robotics, April 2016. [ DOI | .bib | .pdf ]
[4] Nikhil Somani, Alexander Perzylo, Caixia Cai, Markus Rickert, and Alois Knoll. Object detection using boundary representations of primitive shapes. In Proceedings of the IEEE International Conference on Robotics and Biomimetics (ROBIO), Zhuhai, China, December 2015. [ DOI | .bib | .pdf ]
[5] Caixia Cai, Nikhil Somani, Markus Rickert, and Alois Knoll. Prioritized Motion-Force Control of Multi-Constraints for Industrial Manipulators. In Proceedings of the IEEE International Conference on Robotics and Biomimetics (ROBIO), Zhuhai, China, December 2015. [ DOI | .bib | .pdf ]
[6] Nikhil Somani, Caixia Cai, Alexander Perzylo, Markus Rickert, and Alois Knoll. Object recognition using constraints from primitive shape matching. In Proceedings of the International Symposium on Visual Computing (ISVC), pages 783-792. Springer, December 2014. [ DOI | .bib | .pdf ]
[7] Caixia Cai, Emmanuel Dean-Leon, Nikhil Somani, and Alois Knoll. 6D Image-based Visual Servoing for Robot Manipulators with Uncalibrated Stereo Cameras. In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), Sep. 2014. [ DOI | .bib | .pdf ]
[8] Caixia Cai, Nikhil Somani, Suraj Nair, Dario Mendoza, and Alois Knoll. Uncalibrated Stereo Visual Servoing for Manipulators using Virtual Impedance Control. In the 13th International Conference on Control, Automation, Robotics and Vision (ICARCV 2014), Dec. 2014. [ DOI | .bib | .pdf ]
[9] Caixia Cai, Emmanuel Dean-Leon, Dario Mendoza, Nikhil Somani, and Alois Knoll. Uncalibrated 3D Stereo Image-based Dynamic Visual Servoing for Robot Manipulators. In IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), November 2013. [ DOI | .bib | .pdf ]
[10] Caixia Cai, Emmanuel Dean-Leon, Nikhil Somani, and Alois Knoll. 3D Image-based Dynamic Visual Servoing with Uncalibrated Stereo Cameras. In The 44th Internation Symposium on Robotics (ISR), October 2013. [ DOI | .bib | .pdf ]
[11] Nikhil Somani, Emmanuel Dean-Leon, Caixia Cai, and Alois Knoll. Scene Perception and Recognition for Human-Robot Co-Operation. In ICIAP'13 workshop: First International Workshop on Assistive Computer Vision and Robotics (ACVR 2013). The 17th International Conference on Image Analysis and Processing., September 2013. [ .bib | .pdf ]
[12] Nikhil Somani, Emmanuel Dean-Leon, Caixia Cai, and Alois Knoll. Perception and Reasoning for Scene Understanding in Human-Robot Interaction Scenarios. In CAIP'13 workshop: Recognition and Action for Scene understanding (REACTS 2013). The 15th International Conference of Computer Analysis of Images and Patterns., August 2013. [ .bib | .pdf ]
[13] Nikhil Somani, Emmanuel Dean-Leon, Caixia Cai, and Alois Knoll. Scene Perception and Recognition in industrial environments for Human-Robot Interaction. In 9th International Symposium on Visual Computing (ISVC'13). Springer, July 2013. [ DOI | .bib | .pdf ]