Hot-cell facility is a highly shielded tight casing in which highly radioactive substances are remotely handled by manipulators observing the processes through lead-glass windows. The main purpose of this shielding is to provide protection from high radiation of the radioactive materials located inside the cell, resulting in no hazard to personnel. These shielding are designed to be flexible and allow the interiors to be revamped, when necessary, so work can be performed quickly and at the lowest possible cost. The Hot-cells are now a day's area of interest for the industrial robots which can provide assembly tasks and simulations using tele-operations.
Virtual reality (VR) is gaining popularity as an engineering design tool because of intuitive interaction with computer-generated models. The immersive aspects of virtual reality offer more intuitive methods to interact with 3D data than the conventional 2D mouse and keyboard input devices (Peng et al., 2009). In addition, Digital Mock-Up (DMU) is becoming an indispensible tool in streamlining the design to implementation process. Engineers are creating "Virtual" factories, true 3-dimensional environments that allow visualization of the assembly process in much earlier stage (Deidra L. Donald, 1998). In contrast with the physical manipulation, virtual manipulation with many intuitive features are now a days at glance in academic research. For DMU systems with VR technologies, a ship simulator is presented and modeled in virtual reality technique (Xiufeng Z. et al., 2004). The ship maneuvering simulator uses computer virtual reality technique and creates the 3D virtual environment. Berta J. (1999) discussed the functionality required to enable immersive visualization and hands-on interactivity for automobile DMU using a commercial CAD application (CATIA).
Additionally, researchers are trying to reduce the cost of manipulation using virtual environment and allowing a better control to the operators with force and torque feedback. A virtual reality system for maintainability simulation in aeronautics with haptic force feedback is introduced in (Diego et al., 2004), in which a haptic device is used to track hand movements and to return force feedback for providing the sensation of working with a physical mock-up. Similarly Jayaram (Jayaram et al., 1999) have developed a well known VR assembly application called Virtual Assembly Design Environment (VADE) at Washington State University. One or two-handed assembly operations are performed using position tracking and Cyber Glove. The VADE models part behaviors by importing constraints and model data from the CAD packages. Similar research by Wan (Wan et al., 2004) has been conducted at Zhejiang University in creating MIVAS (A Multi-Modal Immersive Virtual Assembly System) and Grasp identification and multi-finger haptic feedback for virtual assembling (Zhu et al., 2004). Until stages, however, most of DMU systems do not include complex robotic manipulators inside virtual environments with haptic interface support and most of robotic simulators (Miller and Allen, 2004; Klingstam and Gullander, 1999) do not allow on-line analyses with real -time virtual reality technologies. . Earlier studies also show that training operations feel more secure and can relate better to the real world process when trained on a simulator with haptic feedback than those trained on a simulator with non haptic feedback (A. BaLijepalli and T. Kesavadas, 2004).
Figure 1 Hot-cell Facility design of a rectangular shielded zone with manipulators and processors
In this research the idea of virtual hot-cell simulator utilizing DMU technology for remote assembly procedures is presented. The simulator is further made-up with the complex robotic manipulations and enhanced haptic feedback interface which results in a training simulator that is real time, haptic guided and has intuitive interface for operator guidance. This simulator is used as a design tool for evaluating Hot-cell operations using virtual prototypes. The simulation was designed for five major uses: visualization, basic concept development of equipment, design verification of manipulators, development and investigation of optimum operation sequence by trajectory recording of the manipulation performed. Furthermore, complex robotic manipulations were performed by implementing real time simulation algorithms. To carry out simulations all the virtual models must follow the same design pattern and characteristics as followed in the real environment design sheets as shown in fig.1. The concept of replacing the physical environment with virtual environments results in significant cost saving for the overall process (Savall et al., 2002). For assembly tasks the over head crane and master-slave manipulators (MSM) are widely used as remote handling devices in hot-cells (J. K. Lee et al., 2006). Optimal assembly involves many individual operations which are manipulated with help of MSM and over head cranes.
2. Simulator Architecture
2.1 Application Interface
The primary goal was to design desktop based application with flexible architecture supporting both the complex robotic motions and digital mock-up support. To control features of this digital mock-up, the models xml based assembly hierarchy is suggested. Prior to simulations the mechanism files from VRML (virtual) models are generated and facility files from the mechanism files are passed to the application to setup the whole environment. For complex robotic manipulations and assembly procedures; real time algorithms are implemented mapping the haptic device interface with the manipulators. The application interface of the Hot-cell simulator is shown in Fig.2 to store the analysis data of these procedures, trajectory recording of facility manipulation during the real time simulation is designed so that it can be exported to file formats which can be later used as pre-process guidance. In addition, features of digital mock-up applications like making parts transparent to see it through or removing the parts that obstruct another part were included for better visibility.
Figure 2 Desktop based Haptic rendered Virtual simulator interface.
2.2 Software Architecture
The Hot-cell simulator is a desktop based application and the data used to manipulate is generated through a haptic interface of virtual environment. This approach offers a number of advantages compared to other methods. The data (consist of velocities, angles, positions, forces and torque) can be extracted and recorded directly, simplifying the data-collection process. In virtual Hot-cell operations the risk of breakdown and breakage of the system is very low; dangerous and costly environments can be analyzed virtually without risks associated with environment.
When selecting the software tools, factors such as cost, ease of programming, and robustness were all taken into consideration. Fig. 3 shows the application configuration architecture of the Hot-cell simulator. The software libraries used to create the application were: Object oriented C++ as the programming language and Microsoft Visual Studio 2005 as the development environment. Object oriented programming concepts were used to combine the functionality of each library. Open Inventor, a library for creating virtual environments, provides the application platform for this research. The Open Inventor API hides many of the lower- level programming details required to develop, test, and debug VR applications Unlike OpenGL. OpenHaptics SDK toolkit (one of the newest generation API) and device drivers from Sensable were used to drive the PHANTOM haptic device (Sensable, 2009). The toolkit allows both lower- level and higher-level programming access to the PHANTOM. A 3D mouse interface was used to control the secondary manipulators such as BDSM (Dual transport manipulator) and over head cranes in Hot-cell for interfacing 3D mouse the Connexion 3D mouse API provided by the manufacturer coded in C++ was used in this application.
Figure 3 software configuration architecture of Hot-cell simulator
2.3 Configuration files and VRML Models
Mostly VRML geometry models are used to setup the virtual environment. These complex geometry files hold the information of the shape and characteristics of the model designed. Complexity increases when all the geometry models along with the virtual environment information which comprises of many models assembled together. To reduce these complex computations a strategy is rendered to compensate the loss of topological information during the translation process of models from CAD to VR systems, the mock-up models are translated as MEC (mechanism) and FAC (facility) files. The MEC and FAC files are written in the XML format. The facility file provides the initial information to the application for building the virtual environment. It includes the information about the position and orientation of the models to be placed in environment and the mechanism files of each model. Fig. 4 shows the sample FAC file for virtual environment.
Figure 4 Facility file of the virtual environment setup with MEC models
Mechanism files are individually built for each model with additional information about the Joints configuration and coordinate frames of the models. The Mechanism file provides the software information about the joint configurations at loading level which are used in calculating the kinematic values in real time simulation of manipulators. Tags in XML are named as frames that follow a hierarchical structure whereby representing the needed information effectively. Fig.5 shows the VRML model and MEC file generated from it that holds following values:
Attribute function such as the type, name and ID.
Physical information such as weight and material.
Display attributes such as model color.
Texture for rending and visualization.
Position and Rotation elements.
Figure 5 Mechanism file for VRML model Generated by the simulator with angle and joints information
When representing data using XML, first a document type definition (DTD) has to be specified. This would govern the data structure contained by the XML file. Tags are in XML and follow a hierarchical structure. The MEC file contains information like the name and ID of the element for visualization and also about joints and the type of joint in the nodes. Besides, the information about the model such as type, dimension, and other parameters, i.e. the relative position and orientation of the feature in the element's local coordinate system are also contained in the MEC file.
2.4 Mapping of Haptic Device
Haptic-rendering process consists of using information received from the virtual environment, evaluating the force and torque to be generated at a given position, velocity, etc. at the operational joint of a haptic interface. The operational joint can be defined as the location on the haptic interface where position, velocity, acceleration, and sometimes forces and torques, are measured.
In order to map a virtual environment with Haptics, the following problems must be addressed (Pearce et al., 1999):
Finding the point of contact: This is the problem of CD (collision detection), which becomes more complex and computationally expensive as the virtual environment becomes more complex.
Generation of contact forces and torques: This creates the "feel" of the object. Contact forces can represent the stiffness of the object, damping, friction, surface texture, etc.
Dynamics of the virtual environment: Objects manipulated in a virtual environment can perform complex moves and may collide with each other.
Computational rate: Computational rate must be high, around 1 kHz or higher, and the latency must be low. Inappropriate values of both of these variables can cause hard surfaces in the virtual environment to feel soft as well as causing system unstable.
Many virtual reality systems have enabled feedback by adding the sense of "Haptic" as one of the interaction methods and this area of research is widely gaining popularity in academia. The ultimate goal is to enhance the realism of environment as shown by various researchers that adding haptic feedback to the virtual environment increases task efficiency (Burdea 1999, Volkov and Vane 2001).
To perform the haptic aided manipulation, mapping of haptic device with the virtual manipulator was necessary. Fig. 6 shows the transformation process used for mapping haptic device coordinates with the simulated virtual manipulator coordinates.
Figure 6 Desktop based Haptic rendered Virtual simulator interface
After loading all the VRML models using MEC and FAC files the application is then launched into two separate threads of Haptics and graphics. The Open HapticsTM toolkit launches a separate high priority and high frequency (~1000 Hz) haptic thread, which is responsible for communicating with the haptic device. Synchronous callbacks are used to take thread safe snapshots of the haptic data for using it in the second thread. The haptic data provides the graphic loop with joints position and orientation information used to display visual feedback and the haptic loop with a force vector used to render force feedback.
Open inventor is used for launching the graphics thread; depending on the performance of the desktop system it can render the entire graphics scene. Each time through the haptic loop, collision detection is performed in Open inventor using it's built in sensor node capabilities to detect intersection of the haptic based manipulator model with other virtual objects. If the sensor intersects an object color of the manipulator end effector is changed and a beef sound is produced to notify the user.
2.5 Design requirement of MSM and BDSM
The master-slave manipulators (MSM) are widely used as a remote handling device in Hot-cell. The slave manipulator shown in fig.7 is attached on the inner wall of the hot cell to maintain and repair the process equipment it is used to handle and transfer the materials (Sung-Hyun Kim et al., 2008). The master manipulator is attached to the outer wall of the hot cell in real environment; whereas, in virtual simulator the haptic device is mapped as master manipulator. The force reflection enables the operator a sense of performing the task by handling a slave manipulator. Furthermore, using this technology complicated operations can be carried out without the risk of damaging or destroying tools or objects. Table 1 shows the working range and the joint information of the MSM used in this simulation.
Figure 7 Comparison of MSM real and Virtual model used in the Simulator
Table 1 Shows the MSM reach
Beside MSM, to cover the area out of the working range of the MSM, a Bridge transported Dual arm Servo-Manipulator (BDSM) is used which is uniquely designed manipulator to cover the unreachable space of MSM (JK Lee et al., 2006). Fig. 8 shows the working range and D-H parameters of the BDSM. The BDSM model consists of four components: a transporter with a telescoping tube set, a slave manipulator and a remote control system. The manipulator has 6 degree of freedom plus a parallel jaw motion and a handle motion. A telescoping tube set which moves the BDSM in a vertical direction sustains the BDSM. The tube set is attached to the trolley-girder system which provides the travel and traverse motions of the BDSM. In this way the BDSM can be located anywhere inside the hot cell. Furthermore, three virtual cameras are mounted at the lintel for monitoring the operation of the BDSM (HJ Lee st al., 2009). Fig. 9 shows the comparison of the Real and the virtual BDSM model. The size and shape of the virtual models for all drawings are matched with the actual ones of the physical MSM and BDSM.
The BDSM slave manipulator is operated by 3d mouse. The 3d mouse is capable of producing translation and rotation vector simultaneously as user pushes, pulls, or rotates the handle designed to control the mouse movement. The translation vector represents the force the user applies to move the handle. The translation vector is fairly easy to interpret. The three components (X, Y, and Z) of the translation vector can be applied in the same manner as similar data from the keyboard or mouse is applied to the viewing transform. The rotations returned from the device represent the vector about which the user is applying a torque.
Figure 8 Head mounted virtual model of BDSM and D-H parameters for BDSM.
Figure 9 Design of real and virtual model of the BDSM
3. ALGORITHM FOR SIMULATIONS
3.1 Algorithm for Main Application
Real time simulation algorithm is built for complex robotic manipulation in the digital mock-up system. The following algorithm for the simulator is developed for virtual assembly task with haptic feedback. Fig. 10 shows the simulation views of the simulator.
Input = Read Haptic device
Output = Updated Virtual Scene
Render the Scene.
Draw bounding boxes for all the virtual models.
Initiate the Haptic interface if 1st time and start reading haptic input.
Detect Collision on Operated Joints of MSM using bounding box information.
(Collision Algorithm) If collision occurs skip step d.
Calculate Inverse Kinematics, Dynamics and manipulate the manipulator.
Provide feedback to the haptic interface.
Update the scene graphics.
If recording is enabled, store active joint information in file.
Until (simulation goals are not achieved)
Figure 10 Simulation views of the simulator processing different manipulations
3.2 Algorithm for Building Bounding Boxes
To check the collision the conventional bounding box approach is followed to mark the boundaries of the models. The algorithm for the bounding boxes generation as shown in Fig. 11 calculates the bounding boxes. The Joint nodes are defined in MEC files for every model with its joint configuration and these configurations are used in calculating bounding boxes. Since only joints are actuated in the simulations the bounding boxes are only built for joint nodes reducing the collision detection complex calculations.
Input = Model Node
Output = Model Node with bounding Box
Get all the Node list of model
If the node type is joint
Build rectangle on the node model
Mark node as "can collide".
Last Node ends in the list
Figure 11 The bounding Boxes generated by the algorithm
3.3 Algorithm for Collision Detection
The algorithm for the collision detection in the simulation is built using the bounding box information calculated for each model during application load stage. The bounding box of a model when obstruct with the bounding boxes of another model the application generates a collision signal. This collision signal is send to the haptic interface that calculates the direction and force required for feed backing. Fig. 12 shows the MSM gripping the bolt and providing the feedback to the operator through haptic interface and graphically changing the color with a sound alert. The algorithm for collision detection is continues thread running during simulation as follows:
Input = bounding boxes of manipulator and the targeted model
Output = collision Beep, Change color
Get cursors bounding boxes (the operated model like MSM, BDSM).
Find from start to end of the target model bounding box
If bounding boxes of manipulator overlapped targeted bounding boxes
Beep sound collision and change the color of the part colliding
Feedback haptic with the collision force
Figure 12 MSM manipulator detecting the collision and feeling the Grip phenomena
4. System Analysis in Virtual environment
4.1 Haptic based manipulation
The manipulation task is initially performed by the human operator through the haptic interface. Since, the feedback forces are kinematically matched with the slave manipulator the operator is able to feel a reflecting force from servo manipulator. The simulator accepts the kinematic models as shown in Fig. 13 hence the user is able to test and analyze typical robot manipulator controls.
Furthermore, the scenario recording architecture embedded in the Hot-cell simulator has been implemented as shown in Fig.14. During simulation the user has an advantage of recording the trajectories being tracked by the manipulator in the scenario. The simulation combines the complex robotic motions performed in a digital mock-up based virtual environment with the haptic based information mapped to simulate the virtual manipulator and its working environment. This system involves real time simulation algorithms and user defined kinematic calculations. The scenario recorded is stored in form of data files. These files can be reused in the virtual environment to assist the trainees or new users to the system; additionally the application provides features to replay, pause and analyze the stored trajectories by zooming into the graphics.
Figure 13 System flow for the manipulation of the MSM and BDSM manipulators
Figure 14 The system architecture of the virtual environment haptic based manipulation.
4.2 Work Space Analysis
In virtual environment the MSM model with position and orientation of the manipulators end effector was analyzed. The voloxidizer comprising: a reaction portion in which spent nuclear fuel is being injected and oxidized was placed in the Hot-cell Scenario of accessing all the joints of voloxidizer was performed and the joints outside the reach of manipulators were identified as shown in Fig. 15. The results of the scenario responded that even the voloxidizer is the reach of manipulator there may be a case in which some parts of it cannot be reached due to the workspace limitations of the manipulator. For this relocation of the equipment was required and a need to deploy BDSM was indentified; the optimal position for BDSM was to be placed in the middle of the facility with its unique design (section 2.5). After deployment of BDSM all the areas of the Hot-cell facility were easily accessible.
Figure 15 The MSM workspace Analysis and voloxidizer relocation.
To make sure the high availability of MSM using digital mock-up virtual environment the workspace of MSM in the Hot-cell was analyzed and optimal layout for the process equipment inside the Hot-cell was obtained. Adapting virtualization in the hot-cell various analyses can be performed to cover the unreachable range of the equipments by placing them or manipulator at optimal location. This virtual information for the re-location of the manipulator and equipments can be effectively used in physical Hot-cell operations enhancing reliability and safety.
4.3 Motor Changing Scenario
To verify the feasibility of maintenance and operations of assembly process a robotic manipulation based graphical simulation was performed. The task for the simulation was to change the motor A (Red) of the voloxidizer with a spare motor B (Blue) already present in the hot-cell. Since the Hot-cell is a shielded facility it was very important to perform these operations with manipulators inside the cell. After performing workspace analysis with motor A in reach of MSM that was to be replaced by Motor B placed at a 2meter distance on the floor. Over head cranes are most widely deployed in the Hot-cell because of their capacity to carry heavy payloads and 4 degrees of freedom to move inside the cell (J. K. Lee et al., 2006). The over head cranes can carry any model from one place to another. To unbolt the voloxidizer safety locks MSM manipulator is operated by haptic interface that provides 6 degrees of freedom input to the application. To lift the motor an overhead Crane is used that is operated by 3D mouse interface that provides 4 degree of freedom input from the user to the application. Motor A was lifted and placed on Ground initially and then Motor B was lifted and placed in Voloxidizer and safety locks were fastened to locked position as shown in Fig. 16 completing the task successfully.
Figure 16. Motor Assembling Scenario in Hot-cell Simulator.
The results of proposed Layout of equipments along with workspace analysis in maintenance process were used which satisfied the tasks in scenario and hence can be effectively used in a physical Hot-cell operations. The designed simulator can be effectively used for remote operations, user training and for analyzing various Hot-cell operations enhancing the reliability and safety issues of environment.
This research investigates the feasibility of a virtual Hot-cell environment for evaluating Hot-cell remote tele-operations. A virtual design of Hot-cell was carried out. Enhanced Haptics interface was implemented for tele-operations. Furthermore, Workspace analysis for the equipments and assembly task by the manipulators were performed and the need for BDSM was identified with a unique design to cover unreached space. The application combines features of real time simulation algorithms, and many software packages including Open Inventor, Open HapticsTM, Visual C++, and haptic force feedback.
After using the application, some conclusions can be drawn: The virtual Hot-cell was designed to assist operators providing more information and guidance during real time tele-operations. Different task scenarios can be simulated and investigated like workspace analysis of manipulators, kinematics modules, haptic feedback, motor replacing, assembly of Voloxidizer and trajectory storing in data files. As expected, this virtual system can be effectively used for real time training and performing robotic manipulation in a Hot-cell. It further provides the benefits of optimizing preprocess assembly design. These remote operations and feasibility analysis performed in virtual simulator effectively saved the cost and provided operators a pre process information and guide.
This work was supported by the Nuclear Research & Development Program of the National Research Foundation of Korea(NRF) funded by the Ministry of Education, Science & Technology(MEST) (Grant code: 2009-0062309).