PS 31-174 - CANCELLED - Coexistence of agents of different systems and the performance of their respective group behaviours based on swarm intelligence

Abstract: Flocking amongst birds serves a number of functions, including improving communication and reducing predation risks. The schooling behaviour of schools of fish helps them to cope with hunting, and after birds form a dense group, eagles will give up hunting them for fear of being hit by the flock. Individuals in biological populations tend to make limited and simple responses but can work together to accomplish collective tasks, which is referred to as swarm intelligence.

Insights from natural systems have inspired work in numerous areas, such as combinatorial optimization, machine learning and image processing. However, the limitations of information sharing, and transfer make it difficult for groups to solve engineering problems in physical space using the swarm intelligence mechanism. While, in nature, birds transfer information to improve the collected data and make decisions without lag during flight. Thus, behavioural modelling of animal swarming may provide new insights into this problem. Starting from this standpoint, we create an intelligent drone swarm that may coexist with other independent systems, such as flocks of birds without affecting the normal life of the other system.

Introducing swarm intelligence to the drone’s model improved the adaptability of drone swarms in real-world tasks and may greatly reduce an individual's demand for environmental information (the natural environment and other individuals within the model) by introducing this collective concept.

For smoother flight and improved controllability, we recorded a drone's flight data (PWM of motors, position, pitch, roll, yaw and thrust), and constructed a virtual drone based on the recorded data. Rather than experimenting in the natural environment, we conducted experiments on virtual drones. By this we avoid damage to the drone or the environment. For implementation, the output was assigned to real-world drones. Experimental results are expected to be consistent with simulation results.

We then draw on real bird behaviours to train the virtual drone swarm to accomplish tasks such as obstacle avoidance and information gathering. Based on the behaviour of birds, our simulated virtual model relies on the integration of information between individuals to improve the quality of environmental data collected. We note that each of the 'virtual drones' has a certain field of view, which not only reduces the memory requirements of the individual but also improves their ability to explore unknown environments (partially known environments). The implementation of this virtual drone modelling approach may reduce the limitations of information in physical space problems.



此页面的语言为英语



翻译为中文（简体）

• 中文（简体）


• 中文（繁体）


• 丹麦语


• 乌克兰语


• 乌尔都语


• 亚美尼亚语


• 俄语


• 保加利亚语


• 克罗地亚语


• 冰岛语


• 加泰罗尼亚语


• 匈牙利语


• 卡纳达语


• 印地语


• 印尼语


• 古吉拉特语


• 哈萨克语


• 土耳其语


• 威尔士语


• 孟加拉语


• 尼泊尔语


• 布尔语(南非荷兰语)


• 希伯来语


• 希腊语


• 库尔德语


• 德语


• 意大利语


• 拉脱维亚语


• 挪威语


• 捷克语


• 斯洛伐克语


• 斯洛文尼亚语


• 旁遮普语


• 日语


• 普什图语


• 毛利语


• 法语


• 波兰语


• 波斯语


• 泰卢固语


• 泰米尔语


• 泰语


• 海地克里奥尔语


• 爱沙尼亚语


• 瑞典语


• 立陶宛语


• 缅甸语


• 罗马尼亚语


• 老挝语


• 芬兰语


• 英语


• 荷兰语


• 萨摩亚语


• 葡萄牙语


• 西班牙语


• 越南语


• 阿塞拜疆语


• 阿姆哈拉语


• 阿尔巴尼亚语


• 阿拉伯语


• 韩语


• 马尔加什语


• 马拉地语


• 马拉雅拉姆语


• 马来语


• 马耳他语


• 高棉语

随时将英语翻译为中文（简体）PRO
一律不翻译英语
一律不翻译www.abstractscorecard.com