Machine Intelligence Competition
A model helicopter which uses sound to navigate has claimed first place in an artificial intelligence competition for a team from De Montfort University Leicester (DMU). This is the second year in a row that DMU has won the top prize in the annual Machine Intelligence Competition, run by the British Computer Society (BCS). Dr Mario Gongora, of the University’s Centre for Computational Intelligence (CCI), and Ben Passow, a PhD student in the CCI, demonstrated the autonomous helicopter during the competition. During the demonstration, the model helicopter had to take off and land in the same spot while flying against a wind generated by a fan. With its automated hearing switched off, the helicopter was unable to find its bearings and drifted off course. With the hearing turned on, it was easily able to fly against the wind and return to its landing position. Ben, who is currently a PhD student at DMU, developed the helicopter as part of his studies. He said: “Researchers working in Artificial Intelligence (AI) have investigated automated vision extensively, but work on automated hearing has been much more limited and has typically been restricted to either the domains of speech and language or fault detection. “Sound offers such a rich source of information if we can provide the machine intelligence to exploit it and a highly complex machine, such as this flying robot can greatly benefit from the information extracted from sound.” He added: “We tend not to notice the vast amounts of information we extract from sound, but sound is all around us all the time. “We can analyse and classify the sounds themselves, their reverberation, diffraction and reflection and if we provide a complex machine with this vast amount of information from sound we can immensely enhance its intelligent behaviour.” Ben presented the helicopter alongside Dr Mario Gongora, of the Centre for Computational Intelligence, who is his PhD supervisor and an expert in computational intelligence. Dr Gongora said: “The helicopter receives information from a supervising computer which extracts that information from the helicopter’s own sound signature. In other words, the helicopter uses its ‘ears’. “This is just the beginning – in future we will be working on extracting individual rotor speeds, servo sounds, and although not tested yet, this system is theoretically capable of working with four helicopters simultaneously.” The system consists of the helicopter and a supervising computer mounted with a number of microphones. This array of microphones receives the helicopter’s intrinsic sound signature and the computer processes this information. Professor Adrian Hopgood, Dean of the Faculty of Technology, said: “The Machine Intelligence Prize is an extremely prestigious award and the standard of entries was outstanding. For a university to win the prize for a second year in succession is unprecedented. “This achievement is testament to both the quality and real-world relevance of our research at DMU.” Ben and Dr Gongora’s presentation, called ‘Fly by Ear’, was chosen as the winning entry by the conference delegates who voted in a secret ballot following live demonstrations by each of the finalists. Their prize was £1,000. The competition took place during the British Computer Society’s 29th SGAI International Conference on Artificial Intelligence (AI-2009), held at Peterhouse College, Cambridge last week.