MIT's SANDO planner can steer a drone around moving obstacles without a prior map, and it comes with a mathematical guarantee of collision avoidance, provided its operating assumptions hold.

It doesn't need to know where obstacles are headed, only a reliable upper limit on how fast they can move. "The only thing the planner needs to know is the top speed the obstacles could reach," says Kota Kondo, who recently earned his doctorate in aeronautics and astronautics at MIT and is lead author of the study in IEEE Transactions on Robotics.

The system detects, groups and tracks moving obstacles, then surrounds each with a virtual sphere showing how far it could travel in any direction over a given time. The further ahead the planner looks, the larger the sphere. A separate heat-map planner steers the drone away from crowded areas, and the onboard computer keeps updating both the corridor and the flight path.

According to MIT, many existing systems offer formal safety guarantees only when obstacles are stationary or known in advance. In simulations, SANDO reached its destination faster than several state-of-the-art comparison systems while avoiding collisions in every environment tested. It also avoided all moving obstacles across 12 flights with a real drone.

The guarantee depends on the obstacle-speed bound holding true. Next steps could include reducing the computational workload and connecting SANDO to machine-learning systems that accept plain-language instructions.

Would you trust a drone like this to fly through a disaster zone?