Dynamic vision sensor to direct lidar scanning

A method comprising:

• generating a point cloud of a scene;
• modeling objects detected in the point cloud in a three-dimensional model of the scene including the objects;
• receiving an event from a dynamic vision sensor that was generated in response to temporal contrast detected at photodetectors of the dynamic vision sensor directed at the scene;
• identifying a region of interest in a field of regard of a lidar system that corresponds to positions of the photodetectors of the dynamic vision sensor that generated the event;
• driving a scanner of the lidar system to perform a scan of the region of interest in the field of regard, the scanner directing emitted light pulses of the lidar along a scan pattern contained within the field of regard of the lidar system;
• tracking the modeled objects in the scene over time using the model, and
• comparing events to tracked objects and preventing driving the scanner to perform a scan of the corresponding region of interest when an event corresponds to a tracked object.

One or more computer-readable non-transitory storage media embodying software that is operable when executed to perform operations comprising:

• generating a point cloud of a scene;
• modeling objects detected in the point cloud in a three-dimensional model of the scene including the objects;
• receiving an event from a dynamic vision sensor that generates the event in response to detecting temporal contrast at photodetectors of a photodetector array directed at the scene;
• identifying a region of interest in a field of regard of a lidar system that corresponds to positions of the photodetectors of the dynamic vision sensor that generated the event;
• driving a scanner of the lidar system to perform a scan of the region of interest in the field of regard, the scanner directing emitted light pulses of the lidar along a scan pattern contained within the field of regard of the lidar system;
• and
• tracking the modeled objects in the scene over time through multiple scan patterns using the model;
• comparing events to tracked objects and not identifying a region of interest if an event corresponds to a tracked object.