Author: Silvano de Gennaro
This document proposes the setup of a preliminary study case upon the developement of a detector maintenence system based on micro-robotics, targeted to all LHC detectors. The system consists of a number of task specific intelligent actuators, ranging from 2 to 8 cm diameter, depending on the task, free to move around the detector along a monorail circuit covering the major critical points that can require maintenence and interventions. Such a system could allow punctual maintenence and diagnostics interventions without the necessity to stop the beam or open the detectors. This will reduce dramatically the number of LHC shutdowns and warmups needed for human intervention.
Teleoperation is not a new concept at CERN. The MANTIS project introduced a teleoperated arm with force feedback, used for occasional interventions. Today, teleoperation has remarkably improved on precision and flexibilty, by using semi-intelligent robots, usually accessed through a VR interface. The operator selects a "virtual tool" to accomplish a certain task. The selected tool will automatically program the robot to perform that task, but the robot can also learn from experience, so the more it performs the task, the better it does it [1]. Applications of this can be found in volcano, deep ocean and space exploration. The nuclear industry is of course another big user of robot technology, mainly for inspection, hazardous handling and maintenance.
Most of these applications use large scale robots, which move on land, water or in space, and perform usually heavy tasks. These robots could not easily fit in our detectors. Also, designing complicated arms to access the detector's inner parts would be a major challange. Fortunately robots do not need to be big. Research in micro-robotics has gone as far as producing an intelligent 1 cube centimeter robot, at EPFL in Lausanne. The "Laboratoire de MicroInformatique" (LAMI) of EPFL developed quite an interesting range of micro-robots. One in particular, the Khepera, developed in collaboration with the Swiss company K-Team SA, features processors powerful enough to be programmed using genetic algorithms, based on neural networks and fuzzy logic. Tests showed that it can reach a good level of autonomous evolutionary learning. The Khepera is based on modular, stackable 55mm diameter boards. Each module contains a processor and is dedicated to a given task. By selecting the appropriate modules, it is possible to configure a robot to accomplish a particular task in certain conditions. For instance, by plugging a radiocontrol turret, a video turret and a grip turret onto Khepera, one has a 70mm tall tobot that can move around, see and detect given kinds of objects, pick them up and put them elsewhere, with a precision of a fraction of a millimiter. One can easily imagine making a Khepera that can, plug and unplug cables, skew trimmers, toggle switches, take electronic measurements and perhaps even weld and replace small boards. Now imagine we put it inside a detector.
Today's technology allows us to perform such operations without even the need to stop the beam, by using intelligent robots.
For reasons of task optimization and miniaturization, it is preferrable to have several different specialized robots instead of a single general purpose one. This approach reduces cost maximising spareability and efficiency. For instance, we can have an Electronic Meter, a Radiation Meter, a Welder, a Cameraman, a Board Plugger, a Cabler, a Gripper, a Tuner etc. Robots can be programmed to collaborate, so that the Cameraman would automatically follow the movement of the Operator (Welder, Cabler etc.), or the Welder would weld-in each wire that the Cabler has plugged, or the Board Plugger would replace a board that the Gripper has removed. The Radiation Meter robot can teach the other robots which areas of the detector are more radioactive, so that they would avoid them when choosing a path to reach a destination. This collaborative strategy features today several implementations in industrial [1], military and research applications [3].