2008 report
Progress: 66%
This highly technical task is divided into three main themes: long-range navigation under sea-ice, short-range communications and satellite telecommunications.
Long-range navigation under sea-ice
The main objective of this work is to develop an under sea-ice localization and navigation system for gliders in DAMOCLES using low frequency sound sources and advanced estimation techniques. The extreme under ice conditions set hard constraints for the design of their navigation systems. The large uncertainty of the system arises not only from the difficult environment, but also from sensors, actuators, internal models and algorithmic approximations the robots use. Normal GPS based localization while on the surface is a rare exception and the system can not be build around it. Additionally, these robots have severe limitations when we talk about their maneuverability. On top of that, payload, energy and finance restrictions make the case even more challenging. The actual work is focused on the detailed conceptual specification of the localisation and navigation system and algorithms for the gliders.
The main progress in the long range navigation task includes:
1. Improvements of the Sterne glider: integration of new attitude sensor (TCM5), new iridium and GPS antenna, temperature and conductivity sensor SBE37 CTD, Benthos 750 Hz hydrophone, clock and TOA modules from Seascan.
2. New functionalities: SOFAR Time of Arrival reception and storage, new pitch control, able to deal with sharp density front, new heading control, based on a predictive approach, routing software using iridium satellite communication, taking into accounts a estimated current model.
3. Sea testing in the Brest area (intensive trials for validating new architecture) and in the Bothnian Gulf, among a triangle of SOFAR Sound sources, in cooperation with FIMR. Sound sources has been moored at a distance of 50km from each other. The glider mission was to navigate among these sources along a 200km trip, listening to SOFAR TOAs, surfacing as soon a TOA was stored and sending by iridium a GPS fix. The aim was to get at sea data from SOFAR sources and reference GPS data in order to be able to evaluate the two algorithms that have been proposed by HUT and ENSIETA.
Based on the work (literature survey, algorithms development, simulator testing and Deliverable D8.1-1) the development of reliable localisation estimation (odometry of the glider + low frequency acoustic information) method and suitable navigation algorithms for various tasks was continued. Work has continued on the batch optimisation algorithm. It processes a window of N latest measurements and optimises simultaneously for the trajectory instead of just the latest pose. The number N can be chosen based on theavailable computational resources.
Algorithm has been adapted for use in embedded processor by minimising memory footprint and computational load. Demanding localisation problem seems to require a 'heavier' algorithm, taking into account a single beacon localisation, more variables to be estimated (f. ex. clock drift or sea current) and a need for greater accuracy. Simulator used for parameter tuning and later mission evaluation needs to know what RAFOS range, what are noise characteristics of range measurements and how accurately can glider movement be calculated without external measurements, not taking into account sea currents. Robustness is the key in autonomous operations thus development needs to concentrate on non-ideal cases, where something goes not as expected. The algorithm must ensure that estimation does not diverge or that we can recover from divergence without human intervention. Robustness must be ensured by comparing several solutions from different algorithms, glider movement avoiding ambiguous configurations, ability to re-initialize an estimation and thorough testing.
The properties of SOFAR signals was be studied during winter 2006-2007 (Tara ice camp). These initial results (similar information also from Craig Lee) revealed that 100 km is most likely the range that can be expected under the ice.
Above information is essential for the development of robust localisation estimation algorithm, which can then be tested extensively, first in a realistic simulator and then later inside a real glider. Additionally, for successful navigation, the following three documents must be defined: Mission Specifications (for gliders), Deployment and Recovery Plan (for gliders, floats and AITPs), and Standard Information Package (from operator via AITP to gliders). For details, please refer to Deliverable D8.1-1.
The further testing of ENSIETA glider was done in order to provide the needed information about the dynamics of the vehicle and about the performance of the on-board sensors. Already achieved under ice glider capabilities include CTD recording, new dive control through sharp density fronts, new heading control. Under ice positioning based on TOAs is in progress. At the surface, the satellite communication (Iridium/GPS) and surface routing based on bathymetry and current were implemented. Envisioned glider missions encompass to reach AITPs performing CTD profile on the way, to cruise to ice free areas, to be able to compute a position fix based on Rafos signals. on-going work on the glider positioning is based on two approaches, using ENSIETA and HUT algorithms which need to be tested on real data. Homing based on positioning (no acoustic homing), which requires knowledge of target absolute position, still needs to be implemented.
12 deployments were performed in the English Channel during Brest trials aimed to validate the new steering and diving capabilities. Fineness control through density fronts, achieved by adjustments by moving a mass based on CTD measurements was validated. Heading control was based on new TCM5 sensor and Kalman filter was applied to estimate a rate of turn.
The Baltic tests in the Bothnian Sea were done with cooperation of ENSIETA, FIMR and UPMC, using sound sources launched with interdistance 50 km for a period October 2007 to January 2008. First attempt to launch Sterne glider done on October 23-26 had to be aborted due to iridium antenna cable failure. On November 22 there was a second series of trials, started with initial weighting and balancing of the glider. Launching was done in bad sea conditions and there was no satellite communication just after launching (rough sea). Iridium message was received after 4 hours of mission but no GPS, next series of Iridium messages were successfully received but without GPS fix. There has been no more contact since October 26 (3,5 significant waves, 5m max) and no TOA were received. Based on this trial it is concluded that a satellite communication antennas should be improved, while this is a critical part of a glider and wing resistance should be increased. Moreover, synchronisation of the SeaScan board should be done off-line and up dated on line.
A new prototype of the STERNE glider will be provided in March 2008 and an integration of positioning algorithm is in progress in both HUT and ENSIETA.
WP 8 Workshop "Integration of acoustics, floats and gliders" took place in Espoo, Finland in May 9-10, 2007. Among participants were Eberhard Fahrbach, Nicolas Seube, Harald Rohr, Tapani Stipa, Janne Paanajärvi, Mika Vainio. The main focus was on glider issues, in particular on status and reality check. The Seaglider (which will be used in Fram Strait) as well as ENSIETA STERNE and DEEP gliders were presented. Strong recommendations (some of them are purely facts) have been worked out for changing several glider related issues, including: the overall timetable, duration of the mission, functionality of the system, etc. etc.
The review of the development of the glider-under-ice programme after the first year of DAMOCLES revealed unexpected difficulties which slowed down the progress in comparison to the DoW. Since it is not acceptable to deploy gliders under ice with a high risk to be lost during a premature operation an alternative plan was developed. It was the aim of the workshop to review the status of the glider programme, to identify the gaps and to develop a modified programme in order to achieve a deployment of gliders under the ice during DAMOCLES. With the way worked out during the meeting we are proceeding in a step by step manner which assures that significant progress towards the objectives of DAMOCLES are achieved without jeopardizing the final goals by accepting to high risks in an early state.
Short range communications
Aquatec Telemetry is the responsible partner in short range communications. A specification of the short range acoustic modems was agreed in 2006 (see deliverable D8.1-4). The latest version of modem specification is now version 2.04 dated 23 May 2007 with additional commands, error codes, etc. Data communication is based on WHOI Micromodem specification with additional custom commands. Electrical specifications covering physical connections, communications interface, power supply, wake-up and mechanical specifications for size and weight were also worked out. Main electronics was designed in 2006 and PCB was commissioned in 2007.
Regarding firmware, the system is using high speed, low power fixed point DSP (Analog Devices Blackfin series). Firmware is written in C for easy editing and modification. The system operates in message-based multi-tasking environment and includes full implementation of the D8.1-4 command set, with serial or USB communications, and data logging to SD card memory. Electromechanical parts include a tuning plate for transducer matching, and commercial acoustic transducer on endcap to suit various housings. PC software, DAMOCLES Acoustic Modem Test Interface, controls modem operation from host PC. Acoustic modeling has been also performed based on ray trace modelling. Early work was completed on a multi-dimensional ray tracing engine operating under Matlab combined with noise and attenuation simulations. The modelling work will be continued.
Regarding communication algorithms, work to date has concentrated on two algorithms: fade tolerant multi-frequency shift keying, using Reed-Solomon encoding and redundant tones (ideally suited to short message transmission) and spread-spectrum based DPSK approach using block encoding for longer burst data. Data rates are currently low (200-500 bps) but with options to test at higher rates depending on range (1-2 kbps feasible).
Equipment delivered to date comprises of:
• Modem board to Scripps for initial integration testing,
• Modem boards to Martec for integration testing,
• 3 x subsea modems and 1x surface modem with 1000m housings and long life battery packs for deployment with Acoustic Tomography Array in Fram Strait.
Data logging and control was tested by Scripps and integration with STAR system was tested in Svalbard.
Resource problems have delayed some areas of development. Modelling is only partially complete – insufficient to test algorithms adequately. Some firmware problems remain to be resolved for future modems, including memory card communication problems and inconsistent transmit and receive sensitivities. Optimised communication algorithms need more work, and critically more testing with both models and in the field. Build problems with other modem boards mean that stock is currently zero.
Work planned for 2008-2009 will be focused on further model development to test algorithms more rigorously. Further trials of existing system will take place: shallow water trials in South of England and deep water trials – e.g. Bergen, or possibly Marseille. Analysis and testing of faster and alternative communication algorithms for optimum data delivery will be performed.
To provide deliverables for 2008-2009 there is necessary to decide how many modems need to be provided, when and in what configuration both for ULS floats and gliders. There is also a question if other modems will be needed. All modems, including those already delivered; require an export license from the UK. If shipped from other EC countries they will also need a license. They may also need a license if re-exported from non-EC countries.
Satellite communication
Two Iridium modems were tested. A complete system for the reception of all the messages coming from various platforms equipped with Iridium modem was developed, in order to ensure the safeguard of the data and to redistribute these data in near real time at the laboratories concerned. Two protocols are available: point to point and Short Burst Data (SBD). Due to the complexity of the point to point protocol, it was suggested to use only the SBD protocol. The server to receive the data (storage and redirection toward the different labs) is ready. Detailed information about the volume and the frequency of data to be transferred daily must be specified by the partners, before this task can be finalized.
Iridium SIM cards were provided for 6 ITPs, one ITAC, one Seaglider and ICE-T bouy. Iridium SIM cards and modems for DAMOCLES users are available from Alain Desautez (alain.desautez@ipev.fr) or Dominique Fleury (dominique.fleury@ipev.fr). Data from POPS are managed on the dedicated IPEV website (see http://www.ipev.fr/damocles. Tracks of five POPS and two ICE-T buoys are available as well raw and decoded data. For each platform state variables (battery power and pressure in operation) can be seen. Meteorological data from POPS (air temperature and barometric pressure) are also available. Data from ICE-T buoys include battery power, stack thermistors values as well as air temperature and barometric pressure. Other data send via IRIDIUM can be included on this web page. DAMOCLES users are asked to provide a specification of the data if relevant.
Plan in 2008 is to define format to link these data to the DAMOCLES data bank
Work to be done during the next 12 months in Task 8.1 (localization and navigation issues) includes:
· Finish the deliverables
· The development of reliable localization estimation algorithms and suitable navigation algorithms continued with a focus on:
- need to know what is RAFOS range extent, what are noise characteristics of range measurements,
- how accurately can glider movement be calculated without external measurements, not taking into account the sea currents,
- need to concentrate on cases where things don't go as expected, non-ideal cases,
- ensuring that estimation does not diverge, or that we can recover from divergence without human intervention.
- comparing several solutions from different algorithms.
- glider movement to avoid ambiguous configurations.
- ability to re-initialise estimation.
· HF modem tests: range, sensitivity, directiviteness
· Simulator improvements
· Real testing with a glider (Brest, Mediterranean, Baltic, Spitsbergen): test, analyse, improve, implement, test ...
However, we definitely need to organise very soon a specialized workshop just for the technical part of WP8. So many issues are still open and this is definitely the last minute to tackle this problem.
