MIT researchers developed a trajectory-planning system designed to help autonomous drones safely navigate unfamiliar environments containing moving and unpredictable obstacles. The system is called SANDO, short for Safe AutoNomous trajectory planning for Dynamic unknOwn environments.
MIT said SANDO can generate efficient flight paths through environments that have not been mapped in advance while still providing a mathematical guarantee that the drone will avoid collisions.
That capability addresses a major limitation in many existing autonomous navigation systems.
Traditional planning approaches can often provide formal safety guarantees when obstacles are stationary or when their locations and future movements are known ahead of time.
Real-world environments are much less predictable.
A drone operating inside a damaged building, around vehicles, near people, or in other dynamic settings may have no prior map and no reliable way to know where an obstacle will appear or how it will move.
SANDO is designed to operate under those conditions.
The planner does not need detailed advance knowledge about the environment or individual obstacles.
Instead, it only needs to know the maximum speed that potential obstacles can reach.
Using that information, the system can continually calculate safe trajectories while the drone moves through the environment.
The approach is intended to allow an autonomous aircraft to respond to changing conditions without sacrificing the formal safety guarantees that are difficult to achieve in dynamic environments.
The research could be particularly useful in applications where drones must operate in places that are dangerous, inaccessible, or rapidly changing.
Potential uses include search-and-rescue missions inside collapsed buildings, where debris may shift and responders or other machines may be moving through the same space.
SANDO could also support wildfire response, allowing autonomous drones to navigate around changing hazards while collecting information or assisting emergency teams.
Other possible applications include mine exploration, where environments may be poorly mapped, and package delivery in crowded areas containing pedestrians, vehicles, and other moving obstacles.
The system could also have broader relevance for autonomous robotics beyond aerial vehicles, particularly in situations where machines must safely operate around unpredictable objects without relying on a complete environmental model.
The work was led by Kota Kondo, who recently completed his doctorate in aeronautics and astronautics at MIT.
The research team also included Jesús Tordesillas and MIT graduate students Juan Rached, Lili Sun, and Yixuan Jia.
Jonathan P. How served as the senior author of the research.
The project reflects broader efforts in autonomous systems research to move robots beyond controlled or highly mapped environments.
By combining real-time trajectory planning with mathematical collision-avoidance guarantees, SANDO is designed to make autonomous navigation more reliable in situations where uncertainty is unavoidable.
KEY QUOTES:
“In the hardest possible environment, where the UAV has no map of the area and there are unknown obstacles moving around, we established a mathematical guarantee of safety. The only thing the planner needs to know is the top speed the obstacles could reach.”
Kota Kondo, Lead Author and MIT Aeronautics and Astronautics Researcher
