Hanczyc Martin M, Parrilla Juan M, Nicholson Arwen, Yanev Kliment, Stoy Kasper
University of Trento and University of Southern Denmark.
Artif Life. 2015 Winter;21(1):47-54. doi: 10.1162/ARTL_a_00151. Epub 2014 Dec 16.
We present a robotic platform based on the open source RepRap 3D printer that can print and maintain chemical artificial life in the form of a dynamic, chemical droplet. The robot uses computer vision, a self-organizing map, and a learning program to automatically categorize the behavior of the droplet that it creates. The robot can then use this categorization to autonomously detect the current state of the droplet and respond. The robot is programmed to visually track the droplet and either inject more chemical fuel to sustain a motile state or introduce a new chemical component that results in a state change (e.g., division). Coupling inexpensive open source hardware with sensing and feedback allows for replicable real-time manipulation and monitoring of nonequilibrium systems that would be otherwise tedious, expensive, and error-prone. This system is a first step towards the practical confluence of chemical, artificial intelligence, and robotic approaches to artificial life.
我们展示了一个基于开源RepRap 3D打印机的机器人平台，该平台能够以动态化学液滴的形式打印并维持化学合成生命。该机器人利用计算机视觉、自组织映射和学习程序，自动对其制造的液滴行为进行分类。然后，机器人可以利用这种分类自主检测液滴的当前状态并做出响应。该机器人被编程为视觉跟踪液滴，要么注入更多化学燃料以维持运动状态，要么引入一种新的化学成分以导致状态变化（例如分裂）。将廉价的开源硬件与传感和反馈相结合，能够对非平衡系统进行可复制的实时操作和监测，否则这些操作将繁琐、昂贵且容易出错。该系统是迈向化学、人工智能和机器人方法在合成生命领域实际融合的第一步。
