Method and system for obstacle avoidance in robot path planning using depth sensors

Machine readable and non-transitory medium having information recorded thereon, where the information, when read by a machine, causes the machine to perform:

• transforming depth data of obstacles, acquired by one or more depth sensors deployed in a 3D robot workspace and represented with respect to a sensor coordinate system, into depth data with respect to a robot coordinate system;
• discretizing the 3D robot workspace to generate a set of 3D grid points representing a discretized 3D robot workspace;
• assigning, based on the depth data with respect to the robot coordinate system, binarized values to at least some of the set of 3D grid points to generate a binarized representation for the obstacles present in the 3D robot workspace;
• determining, with respect to one or more sensing points associated with a part of a robot, whether the part is to collide with any of the obstacles in the 3D robot workspace, wherein the determining further comprises generating a distance map by:
• sub-sampling the set of 3D grid points at a plurality of scales;
• at each scale of the plurality of scales:
• assigning a corresponding maximum-to-compute distance, and
• generating, based on the corresponding maximum-to-compute distance, a distance sub-map with respect to the scale;
• and
• merging distance sub-maps with respect to the plurality of scales;
• and
• planning, based on a result of the determining, a path for the robot to move along while avoiding any of the obstacles.

A system for robot path planning, comprising:

• a coordinate transformer configured for transforming depth data of obstacles, acquired by one or more depth sensors deployed in a 3D robot workspace and represented with respect to a sensor coordinate system, into depth data with respect to a robot coordinate system;
• a workspace discretizing unit configured for discretizing the 3D robot workspace to generate a set of 3D grid points representing a discretized 3D robot workspace;
• a workspace binarizing unit configured for assigning, based on the depth data with respect to the robot coordinate system, binarized values to at least some of the set of 3D grid points to generate a binarized representation for the obstacles present in the 3D robot workspace;
• a collision determining unit configured for determining, with respect to one or more sensing points associated with a part of a robot, whether the part is to collide with any of the obstacles in the 3D robot workspace, the collision determining unit:
• generating a distance map by:
• sub-sampling the set of 3D grid points at a plurality of scales;
• at each scale of the plurality of scales:
• assigning a corresponding maximum-to-compute distance, and
• generating, based on the corresponding maximum-to-compute distance, a distance sub-map with respect to the scale;
• and
• merging distance sub-maps with respect to the plurality of scales;
• and
• a path planning unit configured for planning, based on a result of the determining, a path for the robot to move along while avoiding any of the obstacles.