Please use this identifier to cite or link to this item:
Title: Development of Artificial Intelligence Based Technique for Minimization of Errors and Response time in Head Tracking for Head Worn Systems
Authors: Kataria, Aman
Supervisor: Ghosh, Smarajit
Karar, Vinod
Keywords: Artificial Intelligence;Neural Networks;Prediction;Head Tracking;Head Mounted Display
Issue Date: 9-Nov-2020
Publisher: EIED
Abstract: To overcome the problem of information access in-flight environment, displays such as Head Down Displays (HDD), Head-Up Display (HUD), Helmet Mounted Display (HMD), etc. are incorporated in the cockpits of fighter planes. The motive of the HUDs and the HMDs is to provide the vital information to the pilot of an aircraft so that the pilot need not look down in the cockpit or scan for gathering the flight, weapon, and aircraft information. With the development in head-up displays, the helmet-mounted displays were incorporated to provide the information to the pilot on its visor similar to the case of head-up displays. The information to be displayed on the HMD has to be synchronized with the view of the external environment in synchronism with the head movement of the pilot. For this, the head of the pilot has to be tracked continuously to achieve synchronization with the external environment. For head tracking, various techniques such as mechanical, acoustic, optical, electromagnetic, and inertial can be used. As electromagnetic and optical trackers are more robust, lightweight, accurate, and cost-effective trackers as compared to the other three trackers, therefore, in this work, electromagnetic and optical tracking methods are used for head tracking and validating and predicting the missing data and error estimation. The head tracking devices form a vital component of head-wearable displays. Head tracking devices are used in modern aircraft, especially in combat aircraft. A head tracker updates the position of the head movements of the pilot. As soon as the pilot turns his head, the head tracker passes the orientation data to the controlling computer, which updates the displayed information consequently. Thus, the pilot can get numerous amount of real-time data that is associated with head orientation. To find out the angular coordinates of the pilot’s line of sight, the orientation of the pilot’s head has to be evaluated. The angular orientation of the pilot’s head is comprised of both translational and rotational. The Head tracker provides accurate information to flight computers about the orientation of the pilot’s head with a high degree of accuracy. In this work, six degrees of freedom coordinates of the human head motion consisting of three translational coordinates X, Y, and Z, and three rotational coordinates Yaw, Pitch, and Roll are acquired using the electromagnetic and optical tracker. The 6-DoF coordinates of the head are acquired in the Cockpit Simulator under different environmental conditions such as different light intensities and distance between transmitter and receiver of the tracker. All the experiments are conducted in the cockpit simulation laboratory at CSIR – Central Scientific Instruments Organisation, (CSIR-CSIO), Chandigarh, India. The acquired 6-DoF coordinates of the head are missed due to interferences leading to the 6-DoF missing data set due to the factors like interferences, head of the pilot exceeding the limit of Head Motion Box (HMB), etc. Also, the trackers become more prone to external interferences, ferromagnetism in the case of the electromagnetic tracker, and stray light in the optical tracker. The effect of interferences in the tracking of the head using electromagnetic and optical tracking is also studied in detail. The error in the electromagnetic tracker is calculated with interferences caused by different metals and alloys. Similarly, error in the optical tracker is calculated with interferences caused by different light intensities. Therefore, to avoid the recalibration of the position of the trackers in the cockpit, it is necessary to calibrate the head trackers with respect to the range of HMB. The missing instances of the head coordinates in the 6-DoF data caused by the different kind of interferences is predicted using Self Healing Neural Model, Back Propagation Neural Network, Autoregressive Linear Model and Adaptive Neuro-Fuzzy Inference System. These methods are well-established in predicting the data in such kind of conditions. It is found that the accuracy of prediction is achieved highest using Self Healing Neural Model for both large and small data sets. Accuracy using the Adaptive Neuro-Fuzzy Inference System is higher in small data sets as compared to the large data set. A Self Healing Neural Model is applied to predict the data with 1%, 2%, 10%, 25%, and 35% missed 6-DoF coordinates of the head. The percentage of missing data has been taken in the range of 1% to 35% to simulate the errors due to wide range movement of the head of the pilot beyond the HMB range as well as due to the excessive interference of light in case of optical tracking and magnetic interferences in case of the magnetic tracking method. The results obtained using the Self Healing Neural Model are also compared with other soft computing and regression data predicting techniques. The latency of the electromagnetic and optical trackers may vary depending on the prediction load on the algorithm which again depends on the percentage data missed. Thus, the SHNM algorithm has been developed for predicting the missing data and is then compared with other methods like BPNN, ALM, and ANFIS for accuracy in prediction of missed data due to various factors and latency has been estimated for each of these cases.
Description: Ph.D. Thesis
Appears in Collections:Doctoral Theses@EIED

Files in This Item:
File Description SizeFormat 
Full Thesis with sign.pdfPh.D. Thesis14.25 MBAdobe PDFThumbnail

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.