Three-dimensional motion grid system for autonomous vehicle perception

A method comprising:

• generating for a first frame of a light detection and ranging (LIDAR) point cloud for a vehicle a first grid corresponding to a geographic area around the vehicle, the first grid including a first plurality of cells each having associated therewith an occupancy condition;
• generating for a second frame of the LIDAR point cloud a second grid representing the geographic area around the vehicle, the second grid including a second plurality of cells each having associated therewith an occupancy condition;
• and
• performing a cell-by-cell comparison of the first and second grids to infer motion from relative occupancy conditions of the first plurality of cells and the second plurality of cells.

A control system for a vehicle, the control system comprising:

• a plurality of onboard light ranging and detection (LIDAR) sensors for generating LIDAR data comprising a frame of a point cloud related to an environment of the vehicle;
• and
• a motion grid module comprising a graphics processing unit (GPU) configured to:
• discard points of the point cloud frame that fall outside a defined geographic area around the vehicle;
• subsequent to the discarding, process the remaining points using ray tracing to characterize an occupancy condition of each of a plurality of cells of a three-dimensional (3D) grid corresponding to the defined geographic area, wherein the 3D grid corresponds to the point cloud data frame;
• and
• convert the 3D grid to a two-dimensional (2D) grid, wherein each of a plurality of cells of the 2D grid represents a collective occupancy condition of a portion of the plurality of cells comprising a corresponding column of the 3D grid.