MIT's SANDO plans collision-free drone paths in unknown dynamic environments

MIT researchers developed SANDO, an autonomous navigation system that plans mathematically guaranteed collision-free flight paths for drones in unfamiliar environments with unpredictably moving obstacles. It builds a time-dependent 3D safety corridor and only needs each obstacle's maximum speed rather than exact trajectory predictions. The system was validated on a real UAV in 12 flights with dynamic obstacles.

• SANDO builds a time-dependent 3D safety corridor around possible obstacle movement regions
• It only needs obstacles' maximum speed, not exact trajectory predictions
• In 12 real flights the drone replanned paths onboard without an external map
• Target uses: wildfire response, search and rescue, mines, crowded-area delivery