Dr. Crowthers leads STEM I with a major research project. We start with narrowing down interests in the summer, then spend the first half of the year bringing our independent projects to life. From brainstorming, to reading papers and presenting, Dr. C makes this class more about the process of research than the research itself. Update meetings with Dr. C let us share our progress, working collaboratively with peers for constant feedback. Learning in this class goes beyond books.
The proposed algorithm generates an energy efficient coverage path which enables the search and rescue drone to navigate efficiently through a given area scanning for victims. Moreover, the proposed algorithm enables the usage of multiple drones in a search and rescue mission. The algorithm also has an obstacle avoidance feature which makes enables it to perform in completely unknown environment.
Path planning is an NP-Hard problem used to solve for an optimal path between an initial and final position. While considering the shortest distance, current path planning algorithms do not consider other factors such as energy efficiency, area coverage, or multiple drones necessary for efficient search and rescue missions. A solution to this problem is proposed in this paper. A modified Voronoi partitioning was utilized to divide the area to be searched to ensure equal division of the search area among the different drones. A minimum spanning tree was then constructed for each drone within the partitioned area which was used as the shortest coverage path. To address the energy efficiency problem, 90-degree turns and visiting already searched areas were avoided. After simulating this model in V-REP and taking into account the energy consumption of a drone for a given path, the results yielded that the proposed method had an equal energy consumption to the currently considered most efficient search pattern. However, the coordination abilities of the proposed algorithm are advantageous. Using this algorithm in a search and rescue mission will expedite the process.
Search and rescue missions are crucial in safeguarding human lives during emergencies and natural disasters. These missions often require a substantial investment of resources, including financial funding, fuel, and the dedicated involvement of human rescuers. The complexities in these operations, along with the need for rapid responses, have placed immense strain on organizations involved in search and rescue efforts.
The project's aim is to design a path planning algorithm for an autonomous drone to ensure that the entirety of a given area is scanned, reducing human involvement and resource requirements.
A comparision of the percent error in the area assignment between the proposed algorithm and "DARP" by Kapoutsis et al. 2017.
A comparision of the percent of coverage overlap between the proposed algorithm and "DARP" by Kapoutsis et al. 2017.
A comparision of the energy consumed in a generated path between the proposed algorithm and "DARP" by Kapoutsis et al. 2017.
A path generated by the proposed algorithm to cover a 4x4 grid
A series of statistical tests were conducted to analyze various aspects of the proposed algorithm and compare it with the DARP algorithm. Firstly, a t-test was performed to assess the change in energy consumption relative to obstacle density, revealing a statistically significant lower change in energy consumption for the proposed algorithm (P=0.002). Subsequently, a two-sample t-test indicated a significant difference in mean runtimes between the proposed and DARP algorithms (P=0.0001). Additionally, another two-sample t-test demonstrated that the DARP algorithm exhibited significantly less overlap in coverage (P=0.0001). Finally, a one-way ANOVA was employed, further supporting the conclusion that the DARP algorithm had notably lower overlap in coverage compared to the proposed algorithm outlined in this paper.