Fausto Ferreira

A real-time mosaicking algorithm using binary features for ROVs

This article presents a real-time mosaicking algorithm based on a SLAM framework. The mosaic of the seafloor can be useful in real time for a ROV operator that is piloting the ROV. Two important aspects arise in this kind of work: data association and computational time. In order to solve the first one, a combination of SURF features and template correlation methods is used. To minimize the computational time, a very recent approach in the domain of feature description is used: BRIEF binary features. Finally, to be able to update the whole mosaicking in a fast and easy way, local mosaics are used instead of a global one. The algorithm was tested using data collected in a typical experiment and the results show the improvement with respect to previous versions of a similar algorithm.

Comparison between feature-based and phase correlation methods for ROV vision-based speed estimation

Abstract The performance of different visual approaches for estimating the motion of an underwater Remotely Operated Vehicle (ROV) is discussed. The paper compares three different techniques: feature correlation, Speeded Up Robust Features (SURF), both based on feature extraction and matching, and phase correlation, which instead does not rely on image features. The three algorithms accuracy and performance are compared using a batch of data collected in typical operating conditions with the Romeo ROV. In estimating vehicle speed, phase correlation outperformed SURF in terms of robustness and precision, giving similar results to those obtained with feature correlation. In terms of computational time, phase correlation outperformed both feature-based methods.

Speeded Up Robust Features for vision-based underwater motion estimation and SLAM: comparison with correlation-based techniques

Abstract This paper focus on evaluating the possibility of using Speeded Up Feature Transform (SURF) features (Bay et al. (2006)) as optical features to be tracked in different applications, such as the speed estimation or the SLAM (or the more general dead reckoning) of an underwater Remotely Operated Vehicle (ROV). In particular, SURF-based system performance has been compared to results obtained with a laser-triangulation optical-correlation system using a batch of data collected in typical operating conditions with the Romeo ROV. Although conventional techniques based on automatic extraction of high local variance templates and correlation-based tracking demonstrated their effectiveness in speed estimation, the combination with SURF-based techniques dramatically increases the reliability of the data association process for SLAM and mosaicing applications.

Enhancing autonomous capabilities and human-robot interaction for Unmanned Surface Vehicles

In the context of autonomous exploration and observation of water areas by means of Unmanned Surface Vehicles (USVs), this work describes the improvements developed with respect to the advanced mission control system and the integration with multiple and modular sensing devices, in particular underwater cameras and sonar systems. The experimental proof of the concept validity is obtained testing the overall framework on the CNR-ISSIA Charlie USV. Moreover, to enhance the interaction capabilities between human operator and autonomous platform, different driving and commanding devices, including multi-purpose reconfigurable console and smartphone applications, have been developed and integrated with the already existing architecture. Data gathered from the experimental campaign carried out in Murter (Croatia), within the “Breaking The Surface 2011” training field, are reported.

A survey on real-time motion estimation techniques for underwater robots

Over the last few years, we have assisted to an impressive evolution in the state-of-the-art of feature extraction, description and matching. Feature matching-based methods are among the most popular approaches to the problem of motion estimation. Thus, the need of studying the evolution of the feature matching field arises naturally. The application chosen is the motion estimation of a Remotely Operated Vehicle (ROV). A challenging environment such as an underwater environment is an excellent test bed to evaluate the performance of the several recent developed feature extractors and descriptors. The algorithms were tested using the same open source framework to give a fair assessment of their performance especially in terms of computational time. The various possible combinations of algorithms were compared to an approach developed by the authors that showed good performance in the past. A data set collected by the ROV Romeo in typical operations is used to test the methods. Quantitative results in terms of robustness to noise and computational time are presented and demonstrate that the recent trend of binary features is very promising.

A comparison between different feature-based methods for ROV vision-based speed estimation

Abstract A comparison study between different state-of-the-art visual approaches for estimating the motion of an underwater Remotely Operated Vehicle (ROV) is performed. The paper compares five different techniques: the template correlation, Speeded Up Robust Features (SURF), Scale Invariant Feature Transform (SIFT), Features from Accelerated Segment Test (FAST) and Center Surround Extrema (CenSurE), all based on feature extraction and matching. All these are implemented on the same free open source library which allows a fair comparison that can establish the best technique (depending on the criteria used). Taking into account previous work where SURF and template correlation techniques were evaluated using a batch of data collected in typical operating conditions with the Romeo ROV, the other techniques are compared using the same data set. In estimating vehicle speed, SURF and SIFT presented noise levels higher but close to template correlation, though SURF and SIFT have more outliers. In terms of computational time, template correlation outperforms all other alternatives by large in some cases.

The euRathlon 2015 grand challenge: The first outdoor multi-domain search and rescue robotics competition— A marine perspective

© 2016, Marine Technology Society Journal. All rights reserved. The euRathlon project was an FP7-funded Coordination and Support Action (2013–2015). Itsmain aim was to organize outdoor robotics competitions in realistic search and rescue response scenarios for cooperative land, sea, and air robots. Participant teams were requested to test the intelligence and autonomy of their robots in scenarios inspired by the 2011 Fukushima accident. In the project’s third year euRathlon culminated with the organization of the first outdoor multi-domain search and rescue robotics competition in the world: the euRathlon 2015 Grand Challenge. Sea, air, and land robots were asked to cooperate acting as a robotic intervention team in a scenario simulating an industrial area ravaged by a tsunami. The Grand Challenge was held in Piombino, Italy, in the surroundings of the Tor del Sale power plant, from September 17 to 25. To prepare the teams for the Grand Challenge, two competitions, dedicated to land and marine robots, respectively, took place in 2013 and 2014. In all the competitions, a strong effort was made in benchmarking what led tomeaningful and reasonable scoring principles.Workshops and educational activities complemented the competitions. In this paper, we will focus on the marine robotics competitions of euRathlon with a particular focus on the Grand Challenge. Both technical achievements and general results are presented. The results in terms of team participation and the fruitful effort in dissemination led to establish euRathlon Grand Challenge as the de facto leading search and rescue outdoor robotics competition in Europe.

Improving Automatic Target Recognition with Forward Looking Sonar Mosaics

Abstract Automatic Target Recognition (ATR) is a key element needed to make Mine Countermeasure missions using robots entirely autonomous. While there has been much progress in applying ATR algorithms on high-resolution Synthetic Aperture Sonar (SAS) and sidescan sonar data, performing ATR with a low cost Forward Looking Sonar (FLS) is much more challenging. An algorithm for the detection of underwater man-made objects in FLS previously developed can work in real-time although it suffers considerably from typical noise in sonar images and false alarms. The work presented here shows that ATR algorithms can be exercised on sonar mosaics built also in real-time instead of raw data coming from the FLS. The use of mosaics can help the detection of the targets by reducing some noise (including harmonics from other acoustic devices mounted on the robot) and giving a better contrast to the images to be processed. Moreover, mosaic images can be useful for post-processing and data analysis. The mosaicking algorithm also runs in real-time to maintain the performance of the system and to be useful in real missions. It was tested both on data previously collected and in real experiments with different set-ups and with different sonars. The wide range of results obtained with different surface vehicles and in different situations demonstrate the usefulness of the method.

Comparing region-based and feature-based methods for ROV vision-based motion estimation

Abstract In this work, a comparison between different region-based and feature-based techniques used to estimate the motion of an underwater Remotely Operated Vehicle (ROV) is performed. In what respects region-based detectors, the article compares a previously analyzed template correlation technique with Maximally stable extremal regions (MSER). In previous works, the template correlation method proved to be the best (both in robustness to noise and computational time) when compared with several feature detectors (and descriptors) namely Scale Invariant Feature Transform (SIFT), Speeded Up Robust Features (SURF), Center Surround Extrema (CenSurE), Features from Accelerated Segment Test (FAST), Binary Robust Independent Elementary Features (BRIEF) and Oriented FAST and Rotated BRIEF (ORB). Therefore, the need of comparing it with other region-based detectors arises. Nonetheless, previously untested detectors are now tested combined with BRIEF descriptors due to the good results obtained with BRIEF descriptors in previous works. All the algorithms are implemented in the same free open source library in order to achieve a fair benchmarking, in particular in terms of computational cost. The same experimental data set tested previously is used now in order to allow a relative comparison between these approaches as well as with previous approaches. The qualitative results show that MSER is unsuitable for this application while quantitative results proved that a combination of BRIEF descriptors and a variation of the CenSurE detector is faster and as robust to noise as template correlation. This is an important result as many techniques have been tested so far and all were always slower than template correlation.

An online SLAM-based mosaicking using local maps for ROVs

This article discusses the possibility of building online a mosaic of the seafloor relying on a SLAM framework. The goal is to provide the ROV operator with an approximated seafloors visual map relatively rough. In order to have that map, it is important to get an accurate estimate of the location of the visual landmarks and, in particular, a correct data association when a visual landmark is re-visited by the vehicle. The proposed approach uses the combination of a set of local mosaics constructed in the proximity of the SLAM visual landmarks instead of using a global mosaic. The algorithm was tested using a batch of experimental data in typical operating conditions and the results show the effectiveness of the approach.