The Indian Institute of Technology Madras (IIT-Madras) and researchers from a United States-based university have developed a machine-learning algorithm to protect wildlife. The CombSGPO (Combined Security Game Policy Optimisation) algorithm performs strategic resource allocation and patrolling in green security domains to prevent illegal poaching, logging, and fishing.
According to an IIT-Madras press release, the research team found that a combined and coordinated use of forest rangers and surveillance drones, referred to as resources, was a good way to protect wildlife from poaching. As the resources are limited, the algorithm provides highly efficient, scalable strategies. After the extent of resources has been identified, the algorithm handles resource allocation and strategises patrolling. It uses data from the animal population in the conserved area and assumes that poachers are aware of the patrolling being done at various sites. The drones have object detectors mounted on them to signal and communicate with each other as well as the human patrollers.
The algorithm works on a game theory-based model created by the researchers. Game theory is a theoretical framework for conceiving social situations among competing players. In the context of wildlife protection, game theory pertains to predicting the areas where poaching may take place. These predictions are based on the history of poaching incidents and interactions between poachers and defenders. A researcher said that the game model and the kind of resources they used to simulate such a ‘poaching game’ between the defender (forest rangers and drones) and attackers (poachers) are based on the widely-studied Stackelberg Security Game Model and are linked to drones that have already been deployed in Africa to stop elephant and rhino poaching.
As per the World Wide Fund for Nature (WWF), the wildlife trade poses the second-biggest direct threat to the survival of species after habitat destruction. While several organisations and regulatory authorities are trying to curb the incidences of poaching, the poachers seem to have always remained one step ahead of the patrollers. This collaborative research project is expected to keep poaching incidents in check.
To extend and apply this research in domains like security, search and rescue, and aerial mapping for agriculture, among others, the team is trying to perform sample-efficient, multi-agent reinforcement learning to learn with a limited amount of information since data collection is costly in a real-world scenario.
Apart from protecting wildlife, IIT-Madras has also been developing technology-driven solutions to solve several ecological-based problems that have arisen as a consequence of climate change. Earlier this year, the institute used artificial intelligence (AI) tools to study fuel production from biomass. With increasing environmental concerns associated with petroleum-derived fuels, biomass is a practical solution as a source of energy-dense fuel. In a statement, IIT-Madras explained that the researchers used AI and computer simulation and modelling to understand the concept, which saved time and costs.
The team used a machine-learning method called recurrent neural networks (RNN) to study the reactions that occur during the conversion of biomass into energy-dense syngas (gasification of biomass). As OpenGov Asia reported, the technology is able to predict the composition of the biofuel produced as a function of the time the biomass spends in the reactor. The team used a statistical reactor for accurate data generation, which allows the model to be applied over a wide range of operating conditions. The team believes extracting fuel from biomass could help the country attain fuel self-sufficiency.