Nickels Kevin, Nguyen Hoa, Frasch Duncan, Davison Timothy
Department of Engineering Science, Trinity University, One Trinity Place, San Antonio, TX 78212-7200, USA.
Department of Mathematics, Trinity University, One Trinity Place, San Antonio, TX 78212-7200, USA.
Biomimetics (Basel). 2019 Oct 12;4(4):69. doi: 10.3390/biomimetics4040069.
Mobile robots that can effectively detect chemical effluents could be useful in a variety of situations, such as disaster relief or drug sniffing. Such a robot might mimic biological systems that exhibit chemotaxis, which is movement towards or away from a chemical stimulant in the environment. Some existing robotic exploration algorithms that mimic chemotaxis suffer from the problems of getting stuck in local maxima and becoming "lost", or unable to find the chemical if there is no initial detection. We introduce the use of the RapidCell algorithm for mobile robots exploring regions with potentially detectable chemical concentrations. The RapidCell algorithm mimics the biology behind the biased random walk of () bacteria more closely than traditional chemotaxis algorithms by simulating the chemical signaling pathways interior to the cell. For comparison, we implemented a classical chemotaxis controller and a controller based on RapidCell, then tested them in a variety of simulated and real environments (using phototaxis as a surrogate for chemotaxis). We also added simple obstacle avoidance behavior to explore how it affects the success of the algorithms. Both simulations and experiments showed that the RapidCell controller more fully explored the entire region of detectable chemical when compared with the classical controller. If there is no detectable chemical present, the RapidCell controller performs random walk in a much wider range, hence increasing the chance of encountering the chemical. We also simulated an environment with triple effluent to show that the RapidCell controller avoided being captured by the first encountered peak, which is a common issue for the classical controller. Our study demonstrates that mimicking the adapting sensory system of chemotaxis can help mobile robots to efficiently explore the environment while retaining their sensitivity to the chemical gradient.
能够有效检测化学流出物的移动机器人在各种情况下都可能有用，比如救灾或毒品嗅探。这样的机器人可能会模仿表现出趋化性的生物系统，趋化性是指朝着或远离环境中的化学刺激物移动。一些现有的模仿趋化性的机器人探索算法存在陷入局部最大值和“迷失”的问题，也就是说，如果没有初始检测，就无法找到化学物质。我们介绍了将RapidCell算法用于移动机器人探索具有潜在可检测化学浓度的区域。RapidCell算法比传统趋化性算法更紧密地模仿了（）细菌有偏随机游走背后的生物学原理，它通过模拟细胞内部的化学信号通路来实现。为了进行比较，我们实现了一个经典趋化性控制器和一个基于RapidCell的控制器，然后在各种模拟和真实环境中对它们进行测试（使用光趋性作为趋化性的替代）。我们还添加了简单的避障行为，以探究其如何影响算法的成功率。模拟和实验均表明，与经典控制器相比，RapidCell控制器能更全面地探索可检测化学物质的整个区域。如果不存在可检测的化学物质，RapidCell控制器会在更广泛的范围内进行随机游走，从而增加遇到化学物质的机会。我们还模拟了一个有三重流出物的环境，以表明RapidCell控制器避免了被首次遇到的峰值捕获，而这是经典控制器常见的问题。我们的研究表明，模仿趋化性的适应性传感系统可以帮助移动机器人有效地探索环境，同时保持对化学梯度的敏感性。
