OpenRelief: Clearing the Fog of Disaster

Unmanned aerial vehicle (UAV) technology is becoming increasingly affordable, and when combined with low cost wireless sensor networks and computer vision opens up exciting possibilities for use in disaster relief. Shane Coughlan is co-founder of OpenRelief, an open source project that is taking advantage of these developments in support of building open, modular solutions for disaster relief. In this interview Shane gives us an insight into how the project came about, its aims, the current status and its roadmap.

Can you tell us about the background to OpenRelief

There are two things that lead to OpenRelief:

One was the Tohoku earthquake and tsunami in Japan during March 2011. At that time communications were severely disrupted, situation overviews were very difficult for aid agencies to obtain, and the NGO supply chain inside the disaster zone was subject to a lot of uncertainty. Volunteers - including me - addressed this issue in various ways, including driving around and using GPS to re-map the disaster area. That clearly was not optimal.

The second was the Disaster Relief Panel at LinuxCon Japan 2011, held during June, and naturally covering issues like those faced in Tohoku in great detail. It was obvious that a great deal of work had already been done around areas like high availability and server-side resilience. However, a gentleman present remarked that the Linux guys were not really addressing frontline issues, and that got me thinking.

Thinking lead to talking, and talking like to more intelligent people thinking. That snowballed and lead to where we are today.

What are the specific aims of the project and how will it achieve these?

In a nutshell, OpenRelief is about designing open, modular, information solutions for disaster relief. These solutions can gather critical information for relief workers on the ground, and that will help the right aid to get to the right places at the right time. Because OpenRelief is a community, the goal is to create, test and enable production and use of the solutions. That means we are focused on bringing the solutions to life and encouraging an ecosystem of producers and users to enable actual deployment.

For now we accomplish our goals mainly through networking. We bring people on-board for development from our personal networks, and we aim to kickstart production and deployment in much the same way. Other interested parties are most welcome to lend a hand too. It's really about people from all over the world wanting to work out ways for technology to help save lives in a particular context.

What stage is development at and what does the roadmap look like?

We are currently focused on the development and release of the robot plane and its sensors to investigate and map disaster zones. Later we will develop more tools for situation mapping, restoring communications and coordinating recovery efforts. The goal is to have a range of solutions to support disaster relief.

Development airframes being assembled

When completed the robot will be small enough to be launched from anywhere and smart enough to recognize roads, people and smoke. It will use sensors to measure weather and radiation. The information it collects can then easily be shared with disaster management systems like Sahana Eden.

The first two robot prototypes are about 80% constructed in the UK. The first fully functional robot - in a mechanical sense - should be in Japan for the formal launch of the project at LinuxCon Japan 2012 in June. It's hard to say where the software will be. We probably won't have all the visual recognition in place for June, but will develop that in active testing in the UK and Japan during the June to December period.

The roadmap is focused on all this right now. As soon as the first robot is fully functional we will use our development knowledge to create a longer milestone overview to guide the development of further sensors and systems.

The OpenRelief Roadmap

How many people are currently working on this?

13 people right now, some more active than others, all valued.

Are there any particular skills that the project is short of?

Yes, there are a few areas where we would especially benefit from a helping hand. We need people familiar with airframes and drones to help reduce the learning process in the testing phase, and to help us develop a sensible summary and build system to progress to the production-ready phase. Essentially we have the ability to do testing, but we need to refine everything to ensure that the systems are actually as simple, as study and as useful as possible for disaster relief. It would also be great to have some people familiar with programming the BeagleBoard (especially the DSP) on-board. Refinements and optimization there for visual recognition through OpenCV are a known weakness for OpenRelief at this stage.

A Mock-up of OpenRelief Computer Vision using OpenCV

What are the biggest technological challenges?

I am tempted to say "everything" but on reflection perhaps there are several big picture challenges that we face.

One of the most important is mechanical endurance. The airframes and electronics are going to face a terrible kicking in any real deployment, and that means we are going to have to become rattle, rain and impact resistant.

The precise issues here won't really be uncovered until we are actually doing field tests from July to December, but I expect that we will uncover quite a few critical and challenging problems. It's a theory versus practice thing. Practice always wins.

Above and beyond this, system endurance is another serious challenge. We are using LiPo batteries, and with our current airframe we expect only 15 to 30 minutes range. That variance really depends on how well we balance and optimize the airframe and electronics. Even so, it will translate into a maximum of about 30km in best case scenarios. It's not great but it is the best we can do right now. Moving forward we need to work out a way to put up enduring drones that cover a few hundred kilos at very low cost.

Where we could do with a few wizards right now is on the airframe efficiency front (engine, autopilot). That's probably where we have the most to lose and gain with this iteration of the technology.

How can those interested get involved in development?

Visit www.openrelief.org and join the development, outreach or user mailing list. Everyone is welcome. We need developers, engineers, communication experts, people in mapping, people in computer visuals, GIS and just about everything else. We also need people who can help translate the technical stuff into language that relief workers and aid agencies can easily grok. The idea is to develop tools that are directly applicable and usable in addressing real challenges. That means developing esoteric stuff like instruction manuals and training materials to accompany the insanely cool technology.

How is the project being funded?

Right now by volunteers. For example, I have purchased the first two airframes and the autopilot technology to get the ball rolling. Karl, the co-founder, has also spend money on UK import duties and various bits and bobs. We have also had some technology (like two Nanodes) kindly donated to bootstrap development. When we get to project launch in June we intend to have a non-profit and to solicit donations to help fund prototyping and development.

What is interesting to note is how cheap the actual technology is. For example, the total cost of each robot is about 1,000 USD retail with our current sourcing. The main cost is brainpower and time.

Are there ways that companies can participate?

There are several ways a company can participate. It really depends on the type of company. Let's split them into three categories; component, production and distribution/integration.

Any company producing components that they think might be useful for the project should get in touch and have a chat. We'd be happy to explore prototyping their technology as part of our solutions, and hopefully that can lead to a mutually beneficial situation, whereby we get better solutions and they get expanded sales.

Companies interested in production of OpenRelief compatible drones or sensors are very welcome too. We are a design and testing community and we need to expand our membership to include producers. What we design is Open Source, and that means any company can run with it to address the market.

That brings us to companies involved in distribution or integration. Again, we are looking for all the partners we can get. While OpenRelief can explain to people how the technology works and even help provide material that assists with brainstorming and training, we need active commercial participants to assist relief agencies and workers deploy, maintain and effective use OpenRelief technology in their workflow.

So, in short, any company - as with any developer or other interested party - is very welcome to be part of our community. They can get in touch via our website or just send me a mail at shane@openrelief.org.

What about in-country support — are you looking to build up a network of contacts in particular areas, e.g. those that may be more prone to humanitarian disaster?

The core activity OpenRelief is focused on designing and testing solutions to particular problems regarding communications and mapping. As I said, OpenRelief solutions will gather critical information for relief workers on the ground, and that will help the right aid to get to the right places at the right time.We are looking at fostering a community around this, and it will be stakeholders in that community who ultimately advocate OpenRelief and address individual markets.

Translating that into English, we want contacts in countries where disasters happen. We want them to be part of our community, to help us learn what is needed, and to help inform people how OpenRelief can help solve local problems.

Thank you for your time, Shane!