IoT-based obstacle avoidance and navigation for UGVs in wooded environments using adaptive fuzzy artificial potential field

Abstract

Internet of Things (IoT) applications are increasingly popular, but data collection can be challenging. Unmanned ground vehicles (UGVs) are a practical solution, but navigation control remains difficult. In this study, we develop a framework based on IoT and adaptive fuzzy artificial potential field (AFAPF) for obstacle avoidance and navigation applications in wooded environments. A UGV was deployed in a wooded area with dense obstacles, and light detection and ranging (LiDAR) was used to scan its environment. The proposed IoT-based UGV framework comprises an integrated monitoring platform, an NVIDIA Jetson AGX Xavier, a global positioning system, LiDAR, the AFAPF algorithm, a battery, and a UGV control system; together, these ensure the stable movement of the UGV in unknown environments. In the proposed AFAPF obstacle avoidance method, the distance between the UGV and an obstacle and the density of the LiDAR point cloud representing an obstacle are input to an adaptive fuzzy decision-making method, which adjusts the expansion radius of each obstacle. This enables the UGV to immediately and effectively avoid obstacles. In experiments conducted in two wooded environments unknown by the UGV, the proposed AFAPF method reduced navigation time and driving distance by an average of 17.62 % and 14.87 %, respectively, compared with a comparable nonfuzzy method.