A coordinated army of smart software and pilotless aircraft could help emergency workers save lives and mitigate damage after disaster strikes
IN THE aftermath of a disaster like the massive typhoon Haiyan, which devastated the Philippines on 8 November, confusion often reigns and sketchy information abounds. This can leave responders unsure if their efforts are being put to the best effect.
A coordinated army of smart software and aerial drones could change that. By gathering information from across an affected area, software agents – algorithms that can work with a degree of autonomy – will build a picture of the situation and give recommendations for how people should direct their resources to mitigate damage and save lives.
The system, called Orchid, is being developed as part of a £10 million project of the same name funded by the UK government. Initial testing has shown promising results, and Rescue Global, a London-based disaster responder, is planning a field trial next year (see “Getting Orchid on the ground”).
Orchid’s software agents come in several flavours: they inhabit flying drones with on-board cameras, and servers that sift data coming in from the disaster area, like pictures, tweets or even sensor readings.
Each is programmed to watch for rapid changes to a situation. For example, if air quality sensors suggested that a chemical plant was leaking toxic gas, the sensors could send a signal to drones on a mapping project that could then fly to the scene, take further readings and shoot video from several different camera angles.
The information is then communicated wirelessly to an agent called a planner that assesses it and makes a suggestion to the person coordinating the aid effort on how to proceed.
“We are trying to fix the inefficiencies in deploying emergency responders that prevent proper prioritisation and scheduling of rescue tasks,” says Sarvapali Ramchurn, a computer scientist at the University of Southampton in the UK, which leads a consortium of universities and companies working on the Orchid project.
In the wake of the magnitude 7 quake that devastated Port-au-Prince, Haiti, in January 2010, such a system could have been a huge help, says Ramchurn. “Roads were completely blocked and buildings were down so they had to remap the city to find the accessible zones for relief operations,” he says. That was performed via crowdsourced reports to the Ushahidi website and use of Open Street Map – but it took 48 hours to complete. The Orchid team say software agents in charge of swarms of airborne drones could do that far quicker.
The agents are also designed to address a common problem in disasters: unreliable data. Distressed people in traumatic circumstances can supply inaccurate information. For instance, following the 2011 nuclear disaster at the Fukushima Daiichi power plant in Japan, people bought or built their own Geiger counters to track the spread of radiation. But many of the reported readings were implausibly high.
Orchid’s information-gathering agents weed out reports that appear erroneous, quashing outlying high or low numbers in a dataset. In simulations of radioactive plumes that the team has run, this has worked well.
If it works in the field, Orchid could fill an important role in disaster situations, says Paul Donohue of International Rescue Committee UK in London. “It’s core to where we are going,” he says. “We are always looking to use more technology to try and ensure that we have all the information necessary after a disaster.”