Computer Science Research Institute, Intelligent Systems Research Centre, Ulster University, United Kingdom. Author to whom any correspondence should be addressed.
Bioinspir Biomim. 2017 Jun 5;12(3):035002. doi: 10.1088/1748-3190/aa714c.
The coverage problem consists on computing a path or trajectory for a robot to pass over all the points in some free area and has applications ranging from floor cleaning to demining. Coverage is solved as a planning problem-providing theoretical validation of the solution-or through heuristic techniques which rely on experimental validation. Through a combination of theoretical results and simulations, this paper presents a novel solution to the coverage problem that exploits the chaotic behaviour of a simple biologically inspired motion controller, the Braitenberg vehicle 2b. Although chaos has been used for coverage, our approach has much less restrictive assumptions about the environment and can be implemented using on-board sensors. First, we prove theoretically that this vehicle-a well known model of animal tropotaxis-behaves as a charge in an electro-magnetic field. The motion equations can be reduced to a Hamiltonian system, and, therefore the vehicle follows quasi-periodic or chaotic trajectories, which pass arbitrarily close to any point in the work-space, i.e. it solves the coverage problem. Secondly, through a set of extensive simulations, we show that the trajectories cover regions of bounded workspaces, and full coverage is achieved when the perceptual range of the vehicle is short. We compare the performance of this new approach with different types of random motion controllers in the same bounded environments.
覆盖问题是指计算机器人在某个自由区域内通过所有点的路径或轨迹，其应用范围从地板清洁到排雷。覆盖问题可以通过规划问题来解决，为解决方案提供理论验证，也可以通过依赖实验验证的启发式技术来解决。本文通过理论结果和模拟的结合，提出了一种利用简单的生物启发运动控制器——Braitenberg 车辆 2b 的混沌行为来解决覆盖问题的新方法。虽然混沌已经被用于覆盖问题，但我们的方法对环境的假设限制要少得多，并且可以使用板载传感器来实现。首先，我们从理论上证明了这种车辆——一种众所周知的动物趋触行为模型——在电磁场中表现为电荷。运动方程可以简化为哈密顿系统，因此车辆遵循准周期或混沌轨迹，可以任意接近工作空间中的任何点，即解决了覆盖问题。其次，通过一系列广泛的模拟，我们表明轨迹可以覆盖有限工作空间的区域，并且当车辆的感知范围较短时，可以实现完全覆盖。我们在相同的有限环境中比较了这种新方法与不同类型的随机运动控制器的性能。
