Determining object mobility parameters using an object sequence

A method, comprising:

• receiving a first semantic image, the first semantic image comprising a first object;
• associating at least one first lidar point with the first object;
• determining a predicted location for the first object at a previous point in time based at least in part on the at least one first lidar point;
• identifying a second object of a second semantic image that corresponds to the first object based at least in part on the predicted location of the first object at the previous point in time;
• identifying at least one second lidar point associated with the second object, wherein the at least one second lidar point corresponds to the previous point in time;
• generating a virtual point for the second object using the at least one second lidar point;
• and
• determining at least one mobility parameter of the first object based at least in part on the generated virtual point.

A system, comprising:

• at least one processor configured to:
• receive a first semantic image, the first semantic image comprising a first object;
• associate at least one first lidar point with the first object;
• determine a predicted location for the first object at a previous point in time based at least in part on the at least one first lidar point;
• identify a second object of a second semantic image that corresponds to the first object based at least in part on the predicted location of the first object at the previous point in time;
• identify at least one second lidar point associated with the second object, wherein the at least one second lidar point corresponds to the previous point in time;
• generate a virtual point for the second object using the at least one second lidar point;
• and
• determine at least one mobility parameter of the first object based at least in part on the generated virtual point.

At least one non-transitory storage media storing instructions that, when executed by at least one processor, cause the at least one processor to:

• receive a first semantic image, the first semantic image comprising a first object;
• associate at least one first lidar point with the first object;
• determine a predicted location for the first object at a previous point in time based at least in part on the at least one first lidar point;
• identify a second object of a second semantic image that corresponds to the first object based at least in part on the predicted location of the first object at the previous point in time;
• identify at least one second lidar point associated with the second object, wherein the at least one second lidar point corresponds to the previous point in time;
• generate a virtual point for the second object using the at least one second lidar point;
• and
• determine at least one mobility parameter of the first object based at least in part on the generated virtual point.