SBIR-STTR Award

This project is developing a robot controller that does not have the limitations of previous robot control approaches. It uses a reconfigurable 3-d imaging laser radar (ladar) in the loop control along with a parallel distributed processing system. The parallel transputer distributed processor controls a reconfigurable ladar and a standard 6 degrees of freedom robot arm to provide easily controllable grasping in a space environment. The system has the ability to adaptively grasp objects selected by a human operator. This requires development of algorithms which translate the ladar range and intensity data into grasping commands. The algorithmic task is broken into subtasks distributed among the parallel processors. The system has strong adaptability through a cad reference generator which provides data to correlation algorithms that direct the robot. Path planning uses scene sequences rather than a defined object path developed from a single frame analysis. This allows high system adaptability to time-variable situations in the robot environment. Successful development of this adaptive control system can have significant utility in commercial robotics, manufacturing, and biomedical applications.

Autonomous Technologies proposes a Multi-Dimensional LADAR Vision approach to autonomous satellite servicing and assembly that allows a logical transition from pre-specified (robotic) and manned EVA to vision-assisted telerobotic (man supervised) and vision-dependent autonomous satellite servicing. This is accomplished by relying on direct measurements of registered geometric parameters range, velocity, reflectance images for comparison to the shapes, dynamics, and materials of reference objects in an on-line fashion while adding a registered visible channel to allow operator intervention utilizing familiar displays. As a result of Phase I, the proposed LADAR Vision processing approach will provide six-degree-of-freedom measurement (3 position, 3 orientation) of a single object. The Phase II system will also provide simple object recognition based on hierarchic multidimensional feature matching.