Soils are a complex ecosystem that provides critical services, such as growing food, supplying antibiotics, filtering wastes, and maintaining biodiversity; hence monitoring soil health and domestication is required for sustainable human development. Low-cost and high-resolution soil monitoring systems are challenging to design and build. Compounded by the sheer size of the monitoring area of interest and the variety of biological, chemical, and physical parameters to monitor, naive approaches to adding or scheduling more sensors will suffer from cost and scalability problems. We investigate a multi-robot sensing system integrated with an active learning-based predictive modeling technique. Taking advantage of advances in machine learning, the predictive model allows us to interpolate and predict soil attributes of interest from the data collected by sensors and soil surveys. The system provides high-resolution prediction when the modeling output is calibrated with static land-based sensors. The active learning modeling technique allows our system to be adaptive in data collection strategy for time-varying data fields, utilizing aerial and land robots for new sensor data. We evaluated our approach using numerical experiments with a soil dataset focusing on heavy metal concentration in a flooded area. The experimental results demonstrate that our algorithms can reduce sensor deployment costs via optimized sensing locations and paths while providing high-fidelity data prediction and interpolation. More importantly, the results verify the adapting behavior of the system to the spatial and temporal variations of soil conditions.

The ability to obtain the 3D position of target vehicles is essential to managing and coordinating a multi-robot operation. We investigate an ML-backed object localization and tracking system to estimate the target’s 3D position based on a mono-camera input. The passive vision-only technique provides a robust field awareness in challenging conditions such as GPS-denied or radio-silent environments. Our processing pipeline utilizes a YOLOv5 neural network as the back-end detection module and a temporal filtering technique to improve detection and tracking accuracy. The filtering process effectively removes false positive labels to improve tracking accuracy. We propose a piecewise projection model to predict the target 3D position from the estimated 2D bounding box. Our projection model utilizes the co-plane property of ground vehicles to calculate 2D–3D mapping. Experimental results show that the piecewise model is more accurate than existing methods when the training dataset is not evenly distributed in the sampling space. Our piecewise model outperforms the singular RANSAC-based and the 6DPose methods by 28% in location errors. A less than 10-m error is observed for most near-to-mid-range cases.