Most top drones on the market come with some sort of obstacle avoidance. They have reaction times that are not responsive enough for fast navigation in dynamic environments. This video from UZH Robotics and Perception Group shows how event cameras can be used to enhance drones to enable them to avoid fast moving objects.
As the researchers explain:
approach exploits the temporal information contained in the event stream to distinguish between static and dynamic objects and leverages a fast strategy to generate the motor commands necessary to avoid the approaching obstacles.
The algorithm has a latency of only 3.5ms. The drone tested for this was capable of avoiding multiple obstacles at speeds up to 10m/s indoors and outside.