Autonomous Exploration System for Visual Mapping by Enhancing Path Planning and Exploration Strategy
Achmad Akmal Fikri, Nobutomo Matsunaga · 2024
In terms of autonomous exploration system, optimizing mission success requires improvements in tasks such as environmental modeling, localization, and navigation. While many studies focus on time efficiency and precision in environmental modeling, the importance of path planning in expanding exploration areas is often unconsidered. Therefore, this paper proposes an autonomous exploration system combining a modified real-time rapidly-exploring random trees star (RT-RRT*) algorithm and a centroid-based frontier exploration strategy enhanced by simulated annealing and A* (CFE-SA-A*). Based on the simulation results, our RT-RRT* consistently shows higher accuracy in pose estimation, efficient exploration coverage, and high-quality map generation compared to traditional methods such as A*, online RRT, and online RRT*. Additionally, in many scenarios, the proposed exploration system outperforms other approaches in terms of exploration time, accuracy, and map quality.