Conv1D Energy-Aware Path Planner for Mobile Robots in Unstructured Environments

Driving energy consumption plays a major rolein the navigation of mobile robots in challenging environments,especially if they are left to operate unattended under limitedon-board power. This paper reports on first results of an energyawarepath planner, which can provide estimates of the drivingenergy consumption and energy recovery of a robot traversingcomplex uneven terrains. Energy is estimated over trajectoriesmaking use of a self-supervised learning approach, in which therobot autonomously learns how to correlate perceived terrainpoint clouds to energy consumption and recovery. A novelfeature of the method is the use of 1D convolutional neuralnetwork to analyse the terrain sequentially in the same temporalorder as it would be experienced by the robot when moving. Theperformance of the proposed approach is assessed in simulationover several digital terrain models collected from real naturalscenarios, and is compared with a heuristic inclination-basedenergy model. We show evidence of the benefit of our method toincrease the overall prediction r2 score by 66:8% and to reducethe driving energy consumption over planned paths by 5:5%