Drones, officially known as unmanned aerial vehicles (UAVs), are operated via remote control or automated programs without a human pilot onboard. In recent years, the rapid advancement of advanced sensors and artificial intelligence technology has led to drones being widely utilized across various fields, including delivery, agriculture, disaster relief, and emergency response. Research into autonomous flight technology, which enables drones to perform complex tasks without pilot intervention, is particularly active.
Path Planning Technology for Autonomous Flight
Autonomous flight drones estimate their position and orientation using data collected from multiple sensors. Based on this data, they employ LiDAR and other sensors to create 3D maps to understand surrounding obstacles and terrain. Through this information, drones autonomously generate optimal paths to their target destinations while avoiding barriers and developing flight trajectories that can be realistically followed. Path planning technology is also crucial for operating virtual environment characters and robots, using algorithms to select the best possible route while avoiding collisions.
Key Path Planning Algorithms
There are several key algorithms used to implement path planning. The 'Potential Field' algorithm defines the path from start to goal by using two energies: attractive and repulsive forces. The attractive force draws the object towards the goal, while the repulsive force pushes the object away from surrounding obstacles, helping to avoid them. The Dijkstra algorithm explores all possible paths to calculate the shortest route but has the drawback of high computational demand. The A* algorithm improves this by efficiently calculating the shortest path to the goal with fewer computations.
Other algorithms include Probabilistic Roadmaps (PRM), Rapidly-exploring Random Trees (RRT), and RRT*. The PRM algorithm is helpful in static environments, while RRT can be quickly applied in dynamic settings. RRT* is mainly used in systems where optimal path calculation is crucial.
Recent Trends in Path Planning Research
Recent path planning algorithm developments focus on exploration purposes, such as search and reconnaissance missions, efficiently observing assigned areas. The 'Next-Best-View' (NBV) algorithm allows drones to plan optimal exploration paths while continually scanning unknown environments. Additionally, graph-based and frontier-based exploration planners enable compelling exploration even in complex environments.
Researchers are also exploring path planning for multiple drones rather than single units. The Hong Kong University of Science and Technology is developing algorithms to ensure multiple drones travel along non-overlapping paths. Meanwhile, joint research by MIT and NASA is working on technology to support exploration and rescue missions using multiple drones, even in GPS-denied environments.
The ETRI Autonomous Flight Research Laboratory studies technologies that enable drones to autonomously navigate complex forest environments using SLAM (Simultaneous Localization and Mapping) technology with LiDAR and IMU sensors to generate paths effectively.
As these technologies continue to evolve, the vision of drones thinking and flying independently while delivering goods and responding to disasters is becoming a reality. With continuous and diverse research advancements, the day when autonomous flight capabilities allow drones to coexist harmoniously with humanity is approaching swiftly.
