SUMMARYIn this article, we investigate the problem of parameter identification of spatial–temporal varying processes described by a general nonlinear partial differential equation and validate the feasibility and robustness of the proposed algorithm using a group of coordinated mobile robots equipped with sensors in a realistic diffusion field. Based on the online parameter identification method developed in our previous work using multiple mobile robots, in this article, we first develop a parameterized model that represents the nonlinear spatially distributed field, then develop a parameter identification scheme consisting of a cooperative Kalman filter and recursive least square method. In the experiments, we focus on the diffusion field and consider the realistic scenarios that the diffusion field contains obstacles and hazard zones that the robots should avoid. The identified parameters together with the located source could potentially assist in the reconstruction and monitoring of the field. To validate the proposed methods, we generate a controllable carbon dioxide (CO2) field in our laboratory and build a static CO2 sensor network to measure and calibrate the field. With the reconstructed realistic diffusion field measured by the sensor network, a multi-robot system is developed to perform the parameter identification in the field. The results of simulations and experiments show satisfactory performance and robustness of the proposed algorithms.

中文翻译：

---

现实扩散场中多机器人系统时空变化过程的参数识别

摘要在本文中，我们研究了由一般非线性偏微分方程描述的时空变化过程的参数识别问题，并在现实扩散场中使用一组配备传感器的协调移动机器人验证了所提出算法的可行性和鲁棒性. 基于我们之前开发的使用多移动机器人的在线参数识别方法，在本文中，我们首先开发了一个表示非线性空间分布场的参数化模型，然后开发了一个由协同卡尔曼滤波器和递归最小方法。在实验中，我们关注扩散场，并考虑扩散场包含机器人应避开的障碍物和危险区域的现实场景。确定的参数与定位的源一起可能有助于现场的重建和监测。为了验证所提出的方法，我们产生了一种可控的二氧化碳（CO2) 在我们实验室的现场并建立一个静态 CO2传感器网络来测量和校准现场。利用传感器网络测量重建的真实扩散场，开发了多机器人系统进行现场参数识别。仿真和实验结果表明所提出的算法具有令人满意的性能和鲁棒性。