Researchers from MIT and the University of Pennsylvania have developed a new system that helps drones plan safer and faster flight paths.

The system, named MIGHTY, allows unmanned aerial vehicles (UAVs) to react to obstacles in milliseconds while keeping their flight paths smooth and efficient.

The system uses a new mathematical approach that enables safe and efficient navigation using onboard computers and sensors without relying on expensive proprietary software.

The team claims the model could significantly and reliably improve drone delivery, inspection, and search-and-rescue missions.

Real-time drone planning

In the aftermath of an earthquake, autonomous drones could help rescuers by flying through collapsed buildings to map debris and locate survivors. However, navigating such dangerous and unpredictable environments is a major challenge, as drones must rapidly avoid obstacles while maintaining stable flight.

The MIT and University of Pennsylvania team developed a new trajectory-planning system designed to solve this problem. The technology allows UAVs to react to obstacles in milliseconds while following smooth and efficient flight paths that reduce travel time, making real-time navigation safer and more reliable in complex disaster zones, reports MIT News.

MIGHTY is a new open-source trajectory-planning system designed to help UAVs navigate complex environments safely and efficiently in real time. The system generates smooth, high-quality flight paths while reacting to obstacles using only onboard sensors and computing hardware.

Traditional open-source trajectory planners typically begin by estimating a fixed travel time between two points before calculating the best route. While this approach speeds up computation, it creates limitations in dynamic environments. If a drone is forced to take a longer route to avoid obstacles, it may need to sharply increase its speed to meet the preset travel-time target, thereby reducing its ability to respond safely to sudden hazards.

MIGHTY addresses this problem by optimizing the flight path and travel time simultaneously instead of treating them separately. The system uses a mathematical method known as a Hermite spline to create smooth and precisely controllable trajectories while minimizing travel time.

Efficient trajectory planning

Using both flight path and timing calculations together usually makes the system slower and harder to run in real time.

To solve this, MIGHTY first creates a rough flight path instead of calculating everything from the beginning. It then continuously improves the route step by step using maps generated by the drone’s onboard lidar sensors. This approach reduces the computing workload and allows the UAV to quickly react to obstacles while maintaining a smooth and efficient flight path.

The team claims the approach significantly reduces computational overhead, enabling the UAV to quickly adapt to unknown obstacles while maintaining stable, efficient, and smooth flight trajectories without requiring external computing infrastructure.

In simulation tests, the MIGHTY trajectory-planning system completed tasks using only about 90 percent of the computation time required by leading existing methods while reaching destinations roughly 15 percent faster. The system was also tested on real unmanned aerial vehicles, where it achieved speeds of up to 6.7 meters per second while successfully avoiding all obstacles along its path.

Unlike many trajectory-planning systems that depend on multiple software modules and expensive proprietary solvers, MIGHTY integrates all functions into a single system, enabling faster and more efficient flight planning without external tools. The researchers note that its open-source design makes it easier to deploy in real-world applications such as disaster response, inspection, and autonomous navigation in complex environments, reports MIT News.

Future development plans include expanding the system to coordinate multiple robots simultaneously and conducting more flight experiments in difficult and unpredictable environments while refining the platform through community feedback.