Systems and methods for detecting erroneous lidar data

A non-transitory computer-readable medium for detecting erroneous Light Detection and Ranging (LIDAR) data and storing instructions that, when executed by a processor, cause the processor to:

• receive, from a Light Detection and Ranging (LIDAR) sensor, LIDAR point-cloud data corresponding to a scene in an environment of a robot;
• receive, from a camera, image data corresponding to the scene and generate segmented optical-flow data from the image data;
• fuse, in a two-dimensional (2D) grid, a plurality of objects that include LIDAR points from the LIDAR point-cloud data and optical-flow pixels from the segmented optical-flow data;
• execute a hash function that generates, for the plurality of objects in the 2D grid, a one-dimensional (1D) hash table and an associated index;
• perform one or more queries using the 1D hash table and the associated index to measure an extent of spatial correspondence between the LIDAR points in the 2D grid and the optical-flow pixels in the 2D grid;
• identify the LIDAR point-cloud data as erroneous, when the extent of spatial correspondence fails to satisfy one or more predetermined criteria;
• and
• identify the LIDAR point-cloud data as valid and control operation of the robot based, at least in part, on the LIDAR point-cloud data, when the extent of spatial correspondence satisfies the one or more predetermined criteria.

A method, comprising:

• receiving, from a Light Detection and Ranging (LIDAR) sensor, LIDAR point-cloud data corresponding to a scene in an environment of a robot;
• receiving, from a camera, image data corresponding to the scene and generating segmented optical-flow data from the image data;
• fusing, in a two-dimensional (2D) grid, a plurality of objects that include LIDAR points from the LIDAR point-cloud data and optical-flow pixels from the segmented optical-flow data;
• executing a hash function that generates, for the plurality of objects in the 2D grid, a one-dimensional (1D) hash table and an associated index;
• performing one or more queries using the 1D hash table and the associated index to measure an extent of spatial correspondence between the LIDAR points in the 2D grid and the optical-flow pixels in the 2D grid;
• identifying the LIDAR point-cloud data as erroneous, when the extent of spatial correspondence fails to satisfy one or more predetermined criteria;
• and
• identifying the LIDAR point-cloud data as valid and controlling operation of the robot based, at least in part, on the LIDAR point-cloud data, when the extent of spatial correspondence satisfies the one or more predetermined criteria.