Open-source drone nav avoids obstacles in ms

MIT and the University of Pennsylvania disclosed on May 19 a new open-source drone navigation system called MIGHTY that computes collision-free flight paths in milliseconds using only onboard sensors and compute. The system is designed to let unmanned aerial vehicles replan around obstacles while keeping trajectories smooth and travel times low, according to MIT News and the project paper. The work comes from Kota Kondo of MIT, Yuwei Wu of the University of Pennsylvania, Vijay Kumar of Penn, and Jonathan P. How of MIT. The code is posted publicly on GitHub, and the paper has been accepted to IEEE Robotics and Automation Letters. ### What exactly did the researchers build? MIGHTY is a trajectory planner, not a full drone platform. MIT News said the software generates a feasible path and timing plan together, allowing a UAV to react to obstacles in milliseconds while staying on a smooth path that minimizes travel time. The arXiv paper describes MIGHTY as a Hermite spline-based planner that performs joint spatiotemporal optimization. (news.mit.edu) In plain terms, that means it does not first lock in a route and then separately try to fit a speed profile onto it; it optimizes geometry and timing together. ### Why is the “open-source” part getting attention? MIT News said MIGHTY does not require proprietary software packages that can cost hundreds of thousands of dollars. (news.mit.edu) Kondo said the use of open-source tools means “any researcher, student, or company” can use it freely, and that removing the cost barrier broadens access to high-performance trajectory planning. (arxiv.org) The project repository is public under the MIT Aerospace Controls Laboratory account on GitHub. The repository identifies the paper as accepted to IEEE Robotics and Automation Letters and lists demonstrations including cluttered environments, dynamic obstacles and fast-flight tests. ### What performance did the paper report? The paper reports that, in simulation, MIGHTY reduced computation time by 9.3% and travel time by 13.1% against state-of-the-art baselines, while maintaining a 100% success rate. (news.mit.edu) Those figures come from the authors’ abstract on arXiv. The same abstract said hardware tests included multiple high-speed flights up to 6.7 meters per second in a cluttered static environment, as well as long-duration flights with dynamically added obstacles. (github.com) MIT News separately said the planner is efficient enough for real-time flight using only the robot’s onboard computer and sensors. ### How is this different from older drone planners? Existing open-source planners often trade speed for quality or simplify the problem by fixing travel time early, according to the MIT News report and the paper abstract. (arxiv.org) The researchers said that can force a drone into sharper accelerations or less flexible maneuvers when obstacles appear and the route has to change. Interesting Engineering, citing the MIT work, said MIGHTY uses Hermite splines, initial guesses and lidar-based refinement to speed up joint path-time optimization. That report said the system is aimed at safer, more efficient navigation in dense or changing environments. ### Where could this be used first? (news.mit.edu) MIT News said the researchers see applications in disaster response, last-mile delivery and industrial inspection. The examples it gave included drones moving through collapsed buildings, navigating urban spaces with buildings, wires and people, and inspecting structures such as wind turbines. The next visible step is in the public record already: the code remains on the MIT ACL GitHub repository, and the paper “MIGHTY: Hermite Spline-based Efficient Trajectory Planning” is listed there with its IEEE Robotics and Automation Letters acceptance and DOI. (interestingengineering.com) (github.com) (news.mit.edu)