Energy-orientated path planning for mobile robots using the SMA swarm intelligence algorithm

The need for flexible production and logistics systems is continuously increasing in all industries. Mobile robots can meet these challenges due to their flexibility and range of applications. However, the physical limits of battery capacity lead to uncertainties and a lack of transparency in task processing. This results in low utilization and lower resource usage. Energy-orientated path planning with a prediction of the task-specific energy demand offers a solution for enhancement. Current approaches focus mainly on minimizing the path length. The paper presents a model for energy-oriented path planning of mobile robots based on the Slime Mould Algorithm (SMA). This swarm intelligence algorithm is used to generate optimized paths, minimizing energy demand, and considering obstacles. In addition, the energy and time demand of the paths is predicted based on measurement data from real mobile robots. Firstly, measurement data for the energy demand of different elementary movements of the MiR100 mobile robot is recorded. Based on this, a method for processing the measurement data is developed to determine the necessary parameters for the prediction model. Using the energy demand prediction as a fitness function, the algorithm calculates the required grid points for the paths, focusing on the lowest possible energy demand. The paths are formed by connecting the grid points with linear elements and rounding off the resulting edges with circular elements. To prove the validity and adaptivity, the paths generated by the SMA are measured. The comparison shows that the model accurately predicts the energy and time demand. There is also an improvement compared to internal path planning of the MiR100. The model is also successfully validated using a second mobile robot with different technical characteristics. To make the model accessible, the development is transferred to a public C++ library.