CrowdRank algorithm used  for Searching Flight MH370

CrowdRank algorithm used for Searching Flight MH370

Since the mysterious disappearance of the Boeing 777, with 239 people on board, after leaving Kuala Lumpur for Beijing on March 8, the search for Malaysia Airlines flight MH370 has been blighted by increasing skepticism as the search enters its 13th day. The Search and  Rescue operations now involve 26 countries around the world and over 100 ships, helicopters and airplanes. The costs of this massive operation must easily be running into the millions of dollars. search for flight MH370

Search for Flight MH370 expanded to Northern and Southern Corridor
In addition, investigators are now expanded the searching area and currently looking at two possible corridors – a northern corridor stretching from the border of Kazakhstan and Turkmenistan through to northern Thailand, and a southern corridor stretching from Indonesia to the southern Indian Ocean which covered   2.24 million square nautical miles or 7.68 million square km.

A number of conspiracy theories have arisen after the sudden vanishing such as the plane was turned practically invisible with cutting edge technology. An alternative solution is Crowdsourcing. People power helps turn collection of digital images for searching MH370. “This is a way that ‘people power’ can contribute and help. Crowdsourcing is rapidly emerging as a computing paradigm that can employ the collective intelligence of a distributed human population to solve a wide variety of tasks”, said Dr. Goh Ong Sing, an Associate Professor from Faculty of Information Technology and Communication, Universiti Teknikal Malaysia Melaka (UTeM) who has led research in Artificial Intelligence and headed the First Malaysian Robotic Interactive Programme. This was acknowledged in The Malaysia Book of Record.  Dr Goh and his research students from Malaysia, Jordan and Comoros have joined an effort led by a satellite operator, DigitalGlobe to locate the missing Malaysia Airlines plane, in what may be the largest crowdsourcing project of its kind.goh with his master studentsAssociate Professor Dr Goh Ong Sing, 3rd from left with his research students from Malaysia (Noramirah bt Abdul Rahman & Nurul Ashikin bt Radzid), Comoros (Said Abdou Mfoihaya) and  Jordan (Mu*ath Ibrahim Mohmmad Jarrah & Ahmad Awad Ahmad Husainat) actively engages his students in their real-life learning experience. Today more than 3 million people worldwide have formed a virtual search party as they geo target satellite images from the Gulf of Thailand and now expanded to the Straits of Malacca and as more images are being added daily, including areas in the Indian Ocean. The need for humans or crowd to interpret data streams and this crowdsourcing is the best and fastest way to review thousands of satellite images in the quest to prioritize rescue and recovery efforts for search of the MH370. To help rescue efforts, DigitalGlobe opened its Tomnod platform to the public, essentially delivering a page where visitors can review satellite imagery and tag areas that require attention. The company said more than three million people have participated in the program, with some 257 million “map views” and 2.9 million areas “tagged” by participants. How you can help? A. Join the Tomnod team at  http://www.tomnod.com/ B. As you view the planet imagery and tag what you see

  • Wreckage / Airplane Wreckage
  • Raft / Life Raft
  • Other / Anything interesting or suspicious
  • Oil Slick / Oil Slick in Water

C.  Others on the team mark the same images — you get nods when they agree with you. D. In places where the team agrees, we find the strongest marks

start

More than 3 Million Volunteers join Satellite Search for MH 370 using Tomnod CrowdSourcing

According to Dr Goh, by using its state-of-the-art satellite technology, any member of the public can scroll through 400 sqm patches of sea, scanning for any blemish on the sea’s surface. The hope is that mass public participation might reveal some hitherto unseen clue.  With the availability of the largest group of people possible for a searching boieng MH370. Millions of eyes that are added to those of the military and rescue forces who every day work with ships and aircraft in the same sea. An application developed by Tomnod presented them with squares of imagery where they tagged anything that looked like a raft, wreckage, oil slick or other object of interest or even debris. Tomnod’s internal algorithms take it from there with following steps:

  1. Satellite captures hi-resolution image of Gulf of Thailand and Street of Malacca at half a meter per pixel
  2. Tomnod scales image so it is viewable in browser-based First Insight user interface (UI)
  3. Every mark made by a crowd is automatically, immediately saved to Tomnod. It will record the latitude and longitude of every click from every person and these are compared to determine whether the team agrees.
  4. “The crowd” from everyone in the Tomnod team, agreement emerges at the most interesting locations, feeding data to Tomnod’s machine learning system called CrowdRank, a statistical algorithm, to examine every mark from every person, identify the locations of maximum agreement and figure out what is going on.
  5. Tomnod’s CrowdRank algorithms determine which tagged points have the most consensus among the crowd. The system rankings take into account both which items are most often tagged and which crowd members are most often correct.
  6.  Tomnod delivers a map with crowdsourced tags on the points of interest to DigitalGlobe’s professional analysts. This reduces their workload significantly.

Tomnod’s  algorithm then determines what tags are trustworthy using CrowdRank. It’s supposed to be like Google’s PageRank, but for the crowd. Each point of human analysis is fed into DigitalGlobe’s CrowdRank algorithm to identify areas of crowd consensus and ensure statistical reliability. Every crowd result is filtered by CrowdRank algorithm to locate the highest likelihood sites. CrowdRank score of each location and each member of our crowd was constantly analyzed. CrowdRank is DigitalGlobe’s statistical reliability algorithm that combines the crowd’s inputs to zero-in on the most accurate results. This built-in quality control mechanism is a distinct advantage of using micro-tasking platforms in Tomnod. This project represents an applied research & development initiative. This novel approach combines human collaborative with artificial intelligence, CrowdRank is typically far more accurate at capturing relevant information which is useful for a searching missing flight MH370. The analyzed results and a comprehensive map were made available to first responders and people on the ground in  Keyhole Markup Language (KML)  format which can be downloaded at https://d3rj3tz04ush9z.cloudfront.net/results/malaysiaairsar2014_2014-03-13-22-02.kmz and import  into the 3D Google Earth to see the result. Figures below show the analyzed results of the Gulf of Thailand.

CrowdRank- tags

CrowdRank algorithms determine which tagged points have the most consensuses among the crowd.

CrowdRank-wreckage1

CrowdRank with 21 people confirmed that there is airplane wreckage

Galleries of the search for MH 370 revealed by volunteers using Crowdsourcing 

According to Luke Barrington, CTO and co-founder of Tomnod. “That’s really cool data, and there is load of it being collected, but what we’ve learned from the satellite image companies is that only about 2 percent of this data that they collect every day ever gets looked at by humans. Our ambition at Tomnod is discovering how we can tap into those other pixels, and find value in there and find interesting things.” He added, “That’s the vision, to have the crowd to be the first layer of human insight that then feeds it to the analysts and any one individual is unlikely to get the truth, but as you converge on the wisdom of the crowd, the consensus of multiple and independent individuals, that’s when the truth starts to emerge.” Tomnod was founded as part of collaboration with National Geographic, to help find the lost tomb of Genghis Khan. The site takes its name from the Mongolian word “tomnod,” which means “big eye.” This name makes a lot of sense, as the satellite images used in the service are like having a “big eye” watching over the entire globe. The company has helped review damage after the 2011 earthquake in Christchurch, New Zealand, and it was used by digital volunteers to search for displaced populations in Somalia and has been deployed to search high-resolution satellite imagery for signs of the missing airliner in United States. With latest updates announced by Prime Minister of Australia, DigitalGlobe Inc, confirmed that it had collected satellite images on March 16 that appeared to show debris that may be related to the disappearance of Malaysia Airlines Flight MH370. The images were captured on March 16 by the company’s Worldview-2 satellite at a resolution of about 50 cm, and the company was continuing to collect imagery over the area where the possible debris had been spotted. According to the company,” the efforts of millions of online volunteers around the world allowed us to rule out broad swaths of ocean with some certainty”. Barrington added, Tomnod’s next step was to bring game elements into its crowdsourcing platform, making it a fun environment for volunteers to distill raw data into something more useful. Dr Goh and his postgraduate students from Malaysia and Australia also participated with DeforestAction by EarthWatch Institute, another Crowdsourcing solution to maintain the health and strength of the Borneo rainforest in the face of changes in climate and land use. The rainforests of Borneo are some of the most diverse ecosystems in the world. They are home to thousands of plant species, countless types of insect, a vast array of birds, and some of the world’s most iconic and endangered mammals – the orangutan, Sumatran rhino, clouded leopard, and pygmy elephant.  But rainforests are under constant threat from unsustainable logging practices, farming, and climate change. earthwatch3

EarthWatch, another crowdsourcing solution to maintain the health and strength of the Borneo rainforest in the face of changes in climate and land use

The project is developing the web tools for students and volunteers all over the world to monitor rainforests using updated satellite imagery to provide real time intelligence required to halt illegal deforestation. As a volunteer, you’ll trek deep into the heart of the tropical rainforest and, with a team of scientists and research assistants, assess the structure of the rainforest. You’ll help determine how key plant and animal groups have been impacted by logging and fragmentation, and how this disturbance has affected soil moisture, decomposition, and soil erosion. This information is critical if we are to find ways to conserve Borneo’s remaining forests – and provide a strong scientific foundation for their restoration. According to Edwin Wisse and Eduardo Dias,  “Earthwatchers will be using crowd sourcing with remote sensing to monitor deforestation on Borneo. Participants get a parcel of forest which they can monitor over time. The difference lies in the time scale. In Earthwatchers the participants monitor their piece of forest over a longer time for changes.” Earth Watchers is a ground breaking new software tool to enable young people across the planet to monitor the forests and provide usable intelligence to stop deforestation. It provides a new approach for education by actually involving the students directly in the conservation effort by allowing them to monitor real data and to go beyond tokenistic project / lectures to have a hands on impact. Through Earth Watchers, young people across the planet can expose the illegal deforestation early, allowing local authorities to step in and halt it.  This new transparency and global awareness is made possible by technology developed by Geodan inc from The Netherlands. Here’s how it works. 

  1. Each student is allocated a piece of land to monitor (hexagonal 1.6km2)
  2. Each week a new image is provided by radar satellite data providers allowing students to compare images and look for changes / disturbances.
  3. Students can report disturbances from within Earth Watchers and collaborate with their neighbours, who hold the land around their hexagon, to explore whether or not the disturbance is related illegal activity.
  4. Intelligence is sent directly back to the data center, and illegal activity is reported and investigated in partnership with local authorities.

You can join Earthwatch researchers in the heart of Borneo’s rainforest, where you’ll study the impact of climate change on soil, plants, and wildlife at http://earthwatchers.cloudapp.net/ earthwatch

In Earthwatchers the participants monitor their piece of forest over a longer time for changes.

Crowdsourced solutions data collection and processing are growing in many fields and research areas, including disaster risk management and disaster response. As these solutions are adopted into the mainstream, deployments in conflict and post-conflict settings are also likely to grow. Crowdsourcing has potential to change the reality of civic participation in many developing countries.  Indeed, the potential role of crowdsourcing and interactive mapping in improving aid transparency, effectiveness, and social accountability is significant.

About admin