Multi-sensor Integration System Research Articles

A study on the multisensor integration system. 1st Report, Sensor integration system with fuzzy inference and neural network.

Sensors are important for recognizing the system state environmental status in the intelligent robotic system. Thus, the sensor integration system (SIS) has been studied in a wide range of applications. In this paper, it is shown that the SIS can expand the measurable region of sensors with higher accuracy by multiple sensors and that operators can use the system as easily as a single high-performance sensor system. Systems which have been reported so far do not have flexibility for changing/replacing sensors. Thus, this paper presents an approach to the SIS with the knowledge data base of sensors, so the proposed SIS has the flexibility for changing/replacing sensors. This system consists, of four subsystems: 1) sensors as hardware sensing devices, 2) knowledge data base of sensors (KBS), 3) fuzzy inference, and 4) neural network(NN). This system can estimate the error for the sensor's measured value by fuzzy logical inference with KBS. The measured values are integrated by NN. The inferred error and measured value are put into NN. Then, NN's output gives the integrated value of multiple sensors. Thus, the proposed system is shown to be effective through extensive experiments.

An Integrated System for Land Navigation

Perhaps the most stringent requirement for land navigation is imposed by winter operations in the high arctic. The serious consequences of being lost or failing to find a supply dump in the arctic make reliability and accuracy vital necessities. A combination of circumstances conspire to also make the arctic a particularly difficult area in which to navigate by conventional means. Until GPS becomes fully operational, there will be no single system capable of continuously providing the necessary position and heading accuracy. Even then, reliability considerations will dictate that a self contained or autonomous backup be available, certainly for the military user. The Defence Research Establishment in Ottawa (DREO) has therefore developed a multi-sensor, optimally integrated navigation system to satisfy the present operational requirements of the Canadian land forces, which are most severe in the arctic. This system, called PLANS (Primary Land Arctic Navigation System) was designed to be a highly reliable, high- to moderately-accurate, automatic navigation system for all weather off-the-road use. Although designed primarily to meet an arctic requirement, PLANS would of course be just as applicable for desert navigation, or for any application in which reliability and accuracy are a priority, or where simpler and less costly methods are ineffective. PLANS combines several self contained sensors with two satellite receivers, using Kalman filter techniques to provide a continuous optimal estimate of position, height and heading. Time and attitude are also provided, along with waypointing functions. The self contained sensors consist of a gyrocompass/directional gyro unit, an odometer pickoff for speed, a magnetic fluxgate sensor (for which a detailed geomagnetic field model is used in the software) and a baroaltimeter. The satellite receivers are a single channel C/A-code GPS and a dual channel Transit. This paper describes the requirement, the difficulties in meeting this requirement, the design of a multi-sensor integrated system solution, some simulation results and some field trial results.

A track correlation algorithm for multi-sensor integration

A track correlation algorithm in a multi-sensor integration system for a surveillance mission is proposed. The performance of such track correlation processing is strongly dependent not only on the target state estimation error distribution but also on the target spatial density distribution. Therefore, a track correlation problem is formulated as the likelihood ratio test problem which can take both target state estimation error distribution and target state spatial density distribution into consideration. From this formulation, the correlation algorithm for on-line processing is derived by modifications and approximations. Through analytical evaluations and simulation studies, it is shown that the proposed algorithm is superior to the conventional nearest neighbor algorithm.