SafeTom - Safety by Electrical Capacitance Tomography

Every year, thousands of children are injured and about 100 children are killed by cars due to backover in the United States alone. Approximately 36,000 people are treated in U.S. hospitals every year after incidents with chains saws. These are just two examples. In fact, a considerable part of injuries and fatalities originates from the fact that humans (partially) reside in areas where they just should not be. Safety devices that detect human presence would help to reduce this number and industry is grasping for such technology. However, for many problems of this kind no such technology exists. Furthermore, safety devices will only be applied when they are cheap enough, simple enough and safe enough (in terms of false alarms) such that they can be used without disadvantages. One of the techniques that have the potential to solve such detection problems is capacitive sensing. Several safety systems have been developed at the Institute of Electrical Measurement and Measurement Signal Processing (EMT) in the past two years. While they already show promising performance for certain applications, the technology is not yet fully exploited due to a lack of algorithms and corresponding hardware. In the course of SafeTom methods known from Electrical Capacitance Tomography (ECT) will be investigated for their applicability in safety devices. The specific requirements in safety applications pose many new challenges for this well established technology. The validity of suggested problem solutions will be investigated in SafeTom: The reconstruction must be done in real time (i.e. often in less than 1 ms) in order to trigger protection mechanisms. Therefore, complexity of the algorithms is an important aspect. For this purpose, a novel 2.5D ECT principle will be investigated. It is a must that the sensor system can be easily integrated into existing structures. As a consequence, the space for electrodes is very limited and the region of interest can only be observed from a very limited observation angle, thus the available information is limited compared to common ECT. Here, experimental analysis to develop models of the environments should add sufficient prior knowledge for an appropriate regularization. In safety devices, the sensor is fully exposed to the environment. Therefore, man-made noise and electromagnetic disturbers are important issues and are most relevant for the assessment of theoretical performance bounds. Spread spectrum techniques are suggested in SafeTom.

The electrical capacitance tomography is a technology that allows for determination of the spatial distribution of materials based on their electric properties. Using the so created image, objects in the vicinity of the sensor can be detected and classified. The actual sensors are formed by conductive surfaces, which can be virtually mounted on a variety of surfaces while they almost do not require any space. In addition to the small footprint, these sensors have the advantage that they do not require direct line of sight to the objects, as would be required in optical or acoustic methods. In the present project, it was shown that the Electrical Capacitance Tomography can achieve sufficient accuracy at sufficiently high measurement rate, such that a real-time application is possible. The system can be applied in a wide range of applications and it was successfully demonstrated for robot collision avoidance, classification of objects in a robot grasper and in object detection and classification on a bumper of a car. The developed methods have also been successfully applied in other areas that have originally not been addressed in the project such as ice detection on wind turbines or airplanes. Thus, the project results could actually make some contributions to make the world safer.