Generalized Framework for Quantitative Analysis of Robot Navigation Under Rain Conditions

The deployment of autonomous outdoor robots for complex task execution is steeply increasing amidst technological advancements. This induces a need for effective pre-deployment tools such as simulators and performance analyzers to estimate system behavior under challenging operating conditions. However, the existing performance evaluators often fall short when assessing robots’ capabilities in handling real-world tasks, particularly those involving complex scenarios, especially adverse weather conditions such as rain and exposure to dust. In this work, we present a novel framework for estimating the performance degradation of robots under adverse weather conditions by integrating the sensors related and task related factors. The key components of the proposed framework are the task specific validator and the sensor specific validator, which incorporate performance degradation factors introduced by both task related components and sensor specific components. The validators generate a correlation matrix of performance degradation factors from both sensors and autonomy tasks based on their output. The generated correlation matrix provides the dependencies between the hardware and software components associated with the autonomy of the systems, affecting performance. We evaluated the usability of the proposed framework by conducting experiments with high, moderate, and low intensity rainfalls and estimated performance degradation via the correlation matrix for a 3D lidar based mapping task.