Autonomous exploration planning strategy for a reconnaissance agent

This work outlines a practically realizable yet intelligent, exploration planning strategy developed for guiding a recon agent in an unknown environment with no prior information available. The data acquired on the locale through an off the shelf 2D Laser Range Finder (LRF), has been interpreted as a spatial information spectrum. Utilizing LRFs is a trending method of distance measurement in contrast to ultrasonic sensors with errors like cross talk and specular reflections or power-hungry vision systems. As the LRF readings too contain some noise due to reflectivity problems, methods were developed for preprocessing the measurements. The gathered information spectrum is analyzed through the developed intelligent system to identify the explorable candidate regions and the path planning method has been developed so that the efficacy of exploration is preserved. A novel method of adaptive stepping, which guarantees the optimum exploration capability in environments cluttered with a variety of obstacles, has been introduced, hence, enabling the mobile agent to explore both large open spaces and narrow cluttered spaces without collision. Also, the proposed method of `Intelligent Neighbor Identification' has pushed the information extraction to a maximum while pulling the number of scans and unnecessary traversals to a minimum. The Developed system has been simulated, emulated and tested against complex sceneries. It was implemented on a 4WD mobile platform and tested against real environments. Through a comparison of various scenarios, the robustness of the system was confirmed.