An Intelligent Integrated Navigation and Autopilot System for Uninhabited Surface Vehicles

Global positioning systems (GPSs) are very useful navigational aids for both civilian and military vehicular systems. However, owing to the weakness of the radio signals from the satellites they are susceptible to signal loss when operations are being conducted in confined areas such as rivers, mountains and canyons, and are extremely exposed to being deliberately jammed by terrorists or aggressors during times of extreme tension or conflict. Thus given the vulnerability of GPSs to signal loss, it is prudent not to be totally reliant upon them in the design of navigation subsystems for autonomous vehicles (AVs). One solution to this problem is to enhance the subsystem with an algorithm based on simultaneous localization and mapping (SLAM) techniques. SLAM being the process of simultaneously building a feature based map of the operating environment and utilizing it to estimate the location of an AV. To further improve the capability and performance envelope of an AV it is appropriate to combine a SLAM reinforced navigation subsystem with an adaptive control subsystem thereby transforming them into one fully integrated system.This research proposal aims to design and build a new advanced intelligent integrated navigation and autopilot (IINA) system with adaptive capabilities for uninhabited surface vehicles (USVs). The existing intelligent navigation (IN) subsystem will be complemented with a SLAM algorithm which will be newly designed and also capable of interfacing with other types of more traditional navigation system. The new improved IN subsystem will be benchmarked against the existing navigation subsystem and a navigation system enhanced with an inertial measurement unit supplied by Atlantic Inertial Systems (AIS) who are the industrial collaborator for this project. Whilst the intelligent integrated system will be designed and developed for a marine application, the technology evolved will be able to be transferred and used in other types of AV.A common feature with many SLAM algorithms is the reliance upon extended Kalman filters to act as the information data collection mechanism. In this research proposal an interval Kalman filter (IKF) which uses interval calculus in its design will be employed instead and enhanced using artificial intelligence techniques to construct a fuzzy IKF (FIKF). Scene information extraction for SLAM can be from visual or non-visual sources. Visual SLAM has the benefit of selecting and using robust features gained from video imagery of the local scene. A novel aspect of this work will be the design of a feature-matching algorithm (FMA) that will be capable of operating in night-time conditions. Thus, an information data collection mechanism based on a FIKF will be integrated with the FMA to form the overall SLAM enhancement algorithm.Whilst there are a number of methods for introducing adaptability into an autopilot design, in this research project the approach to be taken will be based on a combination of on-line closed loop identification and model predictive control. Thus the methodology described herein represents a new conceptual framework for the design of marine autopilots. It should also be noted that, to date, all uninhabited marine vehicle system identification trials have been performed in the open loop. Upon the successful completion of the design of the adaptive autopilot, it will be merged with the SLAM enhanced IN subsystem to form the IINA system.

In summary, the research and technological advancements which will be made in this programme of work will result in the following: (1) Current feature trackers perform well on man-made structures in scenes. The highly repetitive textures that are characteristic of natural scenes require a new method to be researched and developed. We propose to assess existing techniques and compare to two new techniques. The ability to establish landmarks in highly difficult natural scenes including low light conditions will open up new opportunities for SLAM operation. The new feature detector will make a significant contribution to the wider machine vision community. (2) The HD video archive will be developed as a public data set to allow other researchers access to this difficult visual domain. Co-lateral data will be made available as well. This will ensure that concurrent with the video stream GPS and gyro data are made available. It is expected this to foster further co-operative work with related machine vision researchers. (3) An attractive alternative Kalman filter design that possesses excellent qualities in terms of adaptability, accuracy, robustness, reliability and stability. (4) A ubiquitously novel SLAM enhancement algorithm which will be capable of enhancing more conventional navigation systems and allow truly autonomous navigation to take place with a night-time operational capability. (5) A new adaptive marine autopilot that will possess a wide operational envelope and be compatible with a range of navigation systems. (6) A new conceptual paradigm for marine autopilot design. (7) An innovative intelligent integrated navigation and autopilot (IINA) system which is technically and practically superior to its predecessors in terms of accuracy, capability and robustness. The control and robotic academic and industrial communities will benefit greatly from the above research outputs. The approaches taken in the design of the Kalman filter, the feature tracking algorithm and the adaptive autopilot will be capable of being converted into design tools for use in a range of different waterborne and land applications. Also both the SLAM algorithm and the IINA system may be modified for usage in land and indoor mobile robots, and underwater robotic systems. Knowledge of the various research outcomes and their potential will be promulgated in number ways including journals and conference proceedings. Of particular note is the maintenance of a dedicated website that will contain all progress reports, trials data, photographs and videos, published papers and software listings, with links provided to/from the websites of co-researchers and related programmes/research centres within the UK, Europe and the rest of the world. The UP Knowledge Exploitation team in collaboration with AIS will create an exploitation proposal which will include a business plan and potential market opportunities. Product(s) based on the above will be commercially exploited in partnership with AIS possibly as stand alone items or as enhancements to their existing product line. Thus a route to the worldwide financially lucrative military and civilian market places for navigation and control systems hardware will be established which will increase the economic competitiveness of the UK in these particular industrial sectors. Clearly the IINA system has the potential for installation in land and waterborne robotic systems where they are required to operate in dense and cluttered environments without the aid of GPS for both military and civilian applications. Additionally, the IINA system by itself could be employed in the marine sector as an automatic control system for ships and large marine craft.