Technical Articles

Simulating Trajectories for Remote Handling Vehicles Carrying Activated Components

By Alberto Vale, Instituto de Plasmas e Fusão Nuclear


A 100-ton vehicle with similar dimensions to an autobus and carrying activated components has to move through a building with a minimum clearance of 30 centimeters between the vehicle and the walls. What is the best path for it to follow? Tasked by ITER with answering that question, my colleagues and I explore motion planning and path following for the vehicles using Trajectory Evaluator and Simulator (TES) software that we built with MATLAB®.

ITER is an international research and development project whose goal is to demonstrate the feasibility of nuclear fusion as an energy source by building a fusion reactor capable of producing 500 megawatts of power from an input of just 10 megawatts. When the reactor becomes operational, workers will be unable to enter the ITER facility to perform essential operations such as the transfer of activated components from the facility’s main Tokamak building to the Hot Cell building for processing, repair, testing, or disposal. Instead, the components will be remotely transported by Cask and Plug Remote Handling System (CPRHS) vehicles, which can weigh up to 100 tons, equipped with autonomous guidance and docking systems (Figure 1).

Figure 1. A cut-out illustration of the Tokamak building with a CPRHS vehicle.
Figure 1. A cut-out illustration of the Tokamak building (left) with a CPRHS vehicle (right).

My colleagues and I at Instituto de Plasmas e Fusão Nuclear (IPFN) are using MATLAB and TES to develop algorithms that define the most efficient trajectory for the CPRHS vehicle and guide it along that trajectory. MATLAB has proven to be an ideal development environment for this work because it enables us to move from idea to prototype much more quickly than with any other tool or programming language available to us.

Modeling the Environment and Vehicles

To identify optimal trajectories and develop control algorithms for the CPRHS vehicles, we needed an accurate model of the facility’s narrow hallways and tight corners. We exported floor plans from the CATIA computer-aided design (CAD) software that engineers were using to design the facility. We used MATLAB and Mapping Toolbox™ to import the data, process it, and create a 2D visualization of the facility (Figure 2). The building designs continued to evolve while we were developing TES, but each time they changed, we simply imported the latest design into MATLAB to keep our model up-to-date. In addition, it is possible to update the map in TES, removing or adding new walls or even creating a new map from scratch.

Figure 2. A 2D map of the Tokamak building, imported into MATLAB.
Figure 2. A 2D map of the Tokamak building, imported into MATLAB.

Next, we developed a MATLAB model of the CPRHS vehicles, including their dimensions and distinctive kinematic properties. The orientation and velocity of each wheel on the vehicle can be controlled independently, enabling the vehicle to pivot, rotate in place, and perform other maneuvers required for navigation in cluttered environments. The wheels can be controlled for line guidance, in which they follow the same path, like an autonomous guided vehicle (AGV), or for free roaming, in which each wheel can be rotated independently (Figure 3).

Figure 3. CPRHS vehicle movement patterns, including line guidance and free roaming.
Figure 3. CPRHS vehicle movement patterns, including line guidance (left) and free roaming (right).

Evaluating and Optimizing Trajectories

Using our MATLAB models of the ITER facilities and vehicles, we developed a motion planning methodology to move the vehicles along optimized trajectories between the Tokamak and Hot Cell buildings. We developed MATLAB algorithms to obtain an initial geometric path using Constrained Delaunay Triangulation (CDT). To optimize this path in terms of clearance and smoothness, we augmented the algorithm with an elastic band approach1 that maximizes obstacle clearance and path smoothness while minimizing path length. We used MATLAB to run simulations and visualize the results as the calculated trajectories were refined (Figure 4).

Figure 4. The line guidance approach applied to the trajectory evaluation for port 16 in level B1 of the Tokamak Building.
Figure 4. The line guidance approach applied to the trajectory evaluation for port 16 in level B1 of the Tokamak Building. From left to right: the initial map with the constrained Delaunay triangulation, the geometrical path, the evaluation of the optimization procedure, and the final trajectory.

We used these simulations to identify potential risks in the cask transport scenarios, evaluate refinements to mitigate these risks, and then export CAD models detailing the proposed changes. For example, we identified corners within the ITER facility that could be made less acute, as well as doors that could be modified to open on their opposite side to reduce the risk of collisions during transport activities. We have improved the algorithm by adding new features, including additional vehicle maneuvers and the identification and consolidation of common paths within different trajectories.

To meet initial ITER requirements, we based our first trajectories on line guidance. Later, we used MATLAB to implement a free roaming approach, which has now been accepted by ITER as an alternative. The free roaming approach applies a Rapidly-exploring Random Tree (RRT) algorithm before using the same elastic band technique to smooth the trajectory (Figure 5).

Figure 5. The free roaming approach applied to the trajectory evaluation for port 2 in level B1 of the Tokamak Building.
Figure 5. The free roaming approach applied to the trajectory evaluation for port 2 in level B1 of the Tokamak Building. From left to right: the initial map, the rapidly exploring random tree, the evaluation of free roaming, and the final trajectory.

We designed the TES to make it easy for researchers to develop, test, debug, and simulate their own trajectory algorithms. TES has a well-defined interface that researchers can use to evaluate new algorithms implemented in MATLAB. A team of researchers in Spain took advantage of this capability to develop a line guidance algorithm using a fast marching square approach. Their implementation proved to be 30% faster than our initial CDT approach, and we have since incorporated it into TES.

Building a Scale Prototype

To test the real-world performance of the calculated trajectories and control algorithms, we built a scale model of the CPRHS vehicle and the Tokamak building (Figure 6). The model vehicle, built with LEGO® components, features exactly the same kinematic capabilities as its 100-ton counterpart. Its wheels are independently actuated by an onboard LEGO MINDSTORMS® NXT processor and remotely guided by Bluetooth®.

Figure 6. The CPRHS vehicle prototype navigating a cardboard model of the Tokamak building.
Figure 6. The CPRHS vehicle prototype navigating a cardboard model of the Tokamak building.

The onboard processor receives commands from a MATLAB control program running on a nearby PC. Researchers can use this program to control the scale vehicle manually or automatically, define simple trajectories, or import optimized trajectories from TES.

In addition to showing how our algorithms and trajectories work in an actual physical environment, this setup enabled us to develop and test localization features planned for the ITER facility. For example, we used MATLAB to acquire data from multiple laser range finders and images from video cameras that we had added to the prototype environment. We developed MATLAB algorithms that use this information to determine the position and orientation of the vehicle in real time and adjust its trajectory to avoid unexpected obstacles.

TES in Graduate Education

In addition to supporting the development of remote handling technologies for ITER, TES is playing an important role in graduate education and research at Instituto Superior Técnico in Portugal. Working with faculty advisors, six master’s students and one doctoral student have published more than 16 papers on trajectory optimization and control algorithm development with TES and MATLAB.  Students have also completed two master’s theses on their work with TES and MATLAB, one that won an award for best master’s thesis on robotics in Portugal and another that received an honorable mention.

We continue to work with students to improve TES. Our group is currently implementing improvements to the localization capabilities and adding support for multiple vehicles. Using MATLAB Compiler® we have created a standalone version of TES that researchers working on the ITER project at other institutions can use to evaluate and simulate trajectory optimization algorithms without installing MATLAB. Lastly, we are extending TES to applications in other fields, such as warehouse management, which will require different kinds of autonomous vehicles operating in vastly different physical environments.

1 Elastic Bands: Connecting Path Planning and Control

About the Author

Alberto Vale is a researcher at Instituto de Plasmas e Fusão Nuclear, at Instituto Superior Técnico, where he works on remote handling technologies for ITER and DEMO. He holds a Ph.D. in electrical engineering from Instituto Superior Técnico, Universidade de Lisboa, and is a member of the Portuguese Board of Engineers (Ordem dos Engenheiros).

Published 2014 - 92157v00

View Articles for Related Capabilities

View Articles for Related Industries