MIT researchers announced a new autonomous navigation system on October 7, 2026, that mathematically guarantees an uncrewed aerial vehicle will avoid collisions with moving obstacles in unmapped environments.
How the trajectory planner works
The system, titled SANDO for Safe AutoNomous trajectory planning for Dynamic unknOwn environments, computes flight paths for uncrewed aerial vehicles encountering unknown moving obstacles. Existing trajectory planners typically rely on static environments or lack formal guarantees that prevent crashes.
According to lead author Kota Kondo, the planner requires only the maximum speed that obstacles could reach. SANDO establishes a time-sensitive safety corridor consisting of connected regions of 3D space that contain no obstacles. The system monitors dynamic obstacles, calculates how far they could travel within a specific timeframe based on their top speed, and places bounding spheres around them to keep the corridor clear.
A heat-map based planner guides the vehicle away from crowded areas with multiple obstacles, while an optimization algorithm determines the fastest path to the target. The vehicle recalculates trajectories onboard to react as surroundings shift.
Testing and validation
In simulation trials, SANDO reached target destinations faster than multiple benchmark systems without colliding with obstacles. The team also conducted 12 test flights using a physical UAV, where the drone successfully avoided all dynamic obstacles using onboard sensors and computing.
The study was authored by Kondo alongside Jesus Tordesillas of Comillas Pontifical University, MIT graduate students Juan Rached, Lili Sun, and Yixuan Jia, and senior author Jonathan P. How of MIT. The findings were published in IEEE Transactions on Robotics.
The researchers noted that future iterations could improve computational efficiency and incorporate machine-learning models to support natural language commands. The work received funding in part from the Defense Science and Technology Agency of Singapore.
Key facts and where they come from
- The planner requires only the maximum potential speed of moving obstacles to establish a mathematical safety guarantee.
The only thing the planner needs to know is the top speed the obstacles could reach.
- The navigation system successfully avoided dynamic obstacles during 12 real-world UAV flight tests.
SANDO also avoided all dynamic obstacles in 12 test flights with a real UAV
- The research paper was published in the journal IEEE Transactions on Robotics.
The research appears in the IEEE Transactions on Robotics.
- The project received partial funding from the Defense Science and Technology Agency of Singapore.
This research is funded, in part, by the Defense Science and Technology Agency of Singapore.